CN105752154B - Vehicle steering control system and method - Google Patents

Abstract

The invention discloses a vehicle steering control method, wherein the vehicle has an actuator for steering the tires of the vehicle. The vehicle steering control method includes: obtaining the position of a target point located at a viewing distance away from the vehicle based on the driving action being performed by the vehicle; predicting the forward position of the vehicle located at a viewing distance away from the vehicle; determining the relationship between the target point position and the forward position of the vehicle. the distance difference between the two; calculate the normalized error by dividing the distance difference by the distance ahead; and determine the steering control command based on the integral of the normalized error. The actuator turns the vehicle's tires according to the steering control instructions to perform different driving actions.

Description

Vehicle Steering Control System and Method

technical field

The present invention relates to a vehicle steering control system and method, enabling the vehicle to automatically perform various maneuvers, including lane keeping, lane changing, left and right turning, and obstacle avoidance.

Background technique

In recent years, new vehicle technologies that relieve drivers from the tiring task of driving and allow vehicles to operate autonomously have been in active development. One of the key technologies is vehicle steering control, which enables the vehicle to automatically perform various driving actions to arrive at a destination from a starting point. Common driving maneuvers include: lane keeping (the vehicle body does not deviate from the lane and remains in the lane); lane change (changing from the driving lane to another lane to maintain the speed of the vehicle, merging, getting off the highway, or preparing to turn left and right) ; turn left and right; and obstacle avoidance (vehicle dodges obstacles in the lane to prevent collision accidents).

In order to achieve these driving actions, common vehicle steering control includes at least two parts: path planning and path following. Path planning means that the vehicle control system first plans or generates an ideal path, and then the vehicle drives according to the path and executes the driving action. Normally in lane keeping mode, the lane followed by the vehicle naturally becomes the ideal path. For other driving actions, the acquisition of the ideal route will be determined according to individual driving actions. For example, the ideal path for a lane change is very different from the ideal path for a left or right turn. In other words, the ideal paths for different driving maneuvers are generated via different models or formulas. In addition, the ideal route is also determined according to other factors, such as vehicle speed. For example, the ideal path for a lane change becomes longer at higher vehicle speeds. The ideal path is usually established by one or more smooth curves, and its mathematical representation methods include splines, polynomials, and so on. When performing driving maneuvers, the system uses the ideal path as a reference for steering control. Therefore, the system needs to store the ideal path before executing the driving action, or continuously generate new ideal paths during the execution of the driving action.

Regarding path following, the system determines steering control commands based on the ideal path and vehicle conditions, and then drives the tires to make the vehicle follow the ideal path. The ways in which steering control commands are determined can be divided into two categories. The first category usually includes: (1) estimating or obtaining the curvature of the ideal path, the current lateral offset of the vehicle relative to the ideal path, and the relative direction angle (heading angle) between the traveling direction of the vehicle and the tangent direction of the ideal path; (2) ) and calculating a steering control command based on a combination of a feed-forward term based on curvature and a feedback term based on the current lateral offset and relative directional angle. The second category usually includes: (1) predicting the possible path of the vehicle based on the state of the vehicle; (2) judging the error between the predicted path and the ideal path for a certain distance ahead; (3) and calculating based on the linear function of the error or the integral of the error Steering control commands.

The current known art concerning vehicle steering control systems suffers from several deficiencies. The first disadvantage is that it is necessary to plan or generate an ideal path in order to meet the needs of steering control as a reference; however, this also increases the complexity of the system and the amount of calculation. Since different driving actions have different ideal paths, the system must store the ideal paths of each different driving actions and all the different driving patterns of the ideal paths. Then, according to the driving action to be performed, find the appropriate driving mode again. The system also needs to use the retrieved driving mode and provide an actual ideal route based on vehicle speed, road shape and other factors. For example, different driving modes can be represented by different sets of equations. Each group corresponds to a type of driving action. Depending on the driving maneuver to be performed, the system retrieves the appropriate set of equations. In addition, according to various factors, such as vehicle speed and road shape, the system further determines different parameters in the retrieved equation set. Finally, the retrieved set of equations and the determined parameters together define the ideal path of the driving maneuver. Moreover, when performing driving actions, if the system is set to continuously update the ideal route, the calculation workload will increase; and if the system is set to store the ideal route, the system must occupy the memory space to perform the stored function , and the system also needs to use the current position of the vehicle to identify the corresponding position in the ideal route. In short, the provision of ideal paths increases the system complexity, as well as the requirements for computing and storage functions.

The second shortcoming is that every type of path following method has deficiencies. In the first type, the steering control command includes a pre-award item and a feedback item. Curvature must be obtained by estimation or pre-stored in digital maps (may result in large capacity or communication requirements, unless the vehicle is only traveling on a limited and pre-set path). In an external environment with rapidly changing curvature, another problem is which curvature should be estimated and used (the curvature of the current position of the vehicle or the curvature of the position in front of the vehicle). The above decision also depends on the consideration of vehicle speed. Moreover, since the source of steering control commands is a combination of forward grant and feedback, the proportion of the two is another issue for performance optimization. When the state of the vehicle and the external environment change, most of the proportions will also change accordingly.

The second type of method has the advantage of being a more straightforward method, that is, without estimating curvature or determining specific gravity; instead, the second type of method needs to predict the path that the vehicle is more likely to travel. Such methods usually use two simple means of forecasting: primary forecasting and secondary forecasting. Under primary prediction, the resulting predicted route is simply a straight line in the direction of the vehicle's heading angle. In other words, the main forecast assumes that the vehicle will continue to maintain its current heading angle for the estimated time frame. The formula for calculating the steering angle command δ is: δ=k·e. Wherein, k is a predetermined proportional constant gain value, and e is an error value, that is, the distance difference between the predicted path and the ideal path at a preset position in front of the vehicle. In terms of secondary prediction, the predicted route is based on the vehicle's current driving direction and usage status. Then, at a preset position in front of the vehicle, the error e between the predicted path and the ideal path is integrated and multiplied by the proportional constant gain value k to obtain the steering angle command: δ=k·(Σe).

Although the second type of method is simpler, it also suffers from serious performance loss. When the ideal path is almost straight, the control system using the primary predictor can achieve satisfactory performance; however, the performance of the control system is very poor when there are sharp curves. When the control system is dominated by secondary predictions, under certain conditions the vehicle can exhibit satisfactory accuracy in following the ideal path. The above conditions include low or medium speeds, and relatively gentle curves. However, under the conditions of the preset distance ahead (constant) and the predetermined gain value (constant), when the vehicle is driving at high speed, the control system will become unstable and cause the vehicle to deviate from the path when driving a very sharp curve .

In order to solve the above problems, some prior art uses multiple controllers and switches among them. In one prior art, three controllers are used. The first controller calculates the error using a linear function by means of the primary prediction. The second and third controllers integrate errors by means of secondary predictions. The difference between the second and third controllers is that the second controller uses a smaller predetermined (constant) distance ahead (closer distance), while the third controller uses a larger predetermined (constant) Distance ahead (longer distance). When the vehicle is driving on a straight road, the control system uses the first controller. When the vehicle is driving on very tight curves, the control system uses the second or third controller. However, this technology requires switching between different controllers, which also leads to sacrificing control fluency when switching. If additional management mechanisms are implemented in order to improve the smoothness of controller switching, the system will become more complex. In addition, the technology must ensure that the controller switching decision selects the correct controller in all cases.

Therefore, for relevant personnel, it is urgent to develop a steering control method and system to solve the above-mentioned problems related to path planning and path following faced by the prior art. The invention provides such a steering control method and system.

Contents of the invention

One embodiment of the present invention provides a vehicle steering control method, so that the vehicle can perform different driving actions. The vehicle is equipped with actuators to rotate the tires of the vehicle. The steering control method includes: obtaining the position of the target point at a look-ahead distance of the vehicle; predicting the forward position of the vehicle at the position of the look-ahead distance; determining the distance difference between the target point and the forward position; dividing by the look-ahead distance to calculate a normalized error; and determining a steering control command based on the integral of the normalized error. The steering control command can be determined by multiplying the integral of the error by a gain (fixed or dynamic). Then, according to the steering control command, the actuator turns the tires of the vehicle to perform different driving actions.

In the steering control method, the look-ahead distance is determined by at least one of the following: vehicle speed, vehicle yaw velocity, vehicle lateral acceleration, vehicle steering angle, lane curvature, normalization error, distance difference between target point and forward position, The distance between the vehicle and the lane line when the vehicle is following the lane, the driving action performed by the vehicle, and the position of the obstacle when the obstacle avoidance is performed. In one embodiment, the look-ahead distance is a linear function of vehicle speed. That is to say, when the speed of the vehicle increases, the viewing distance also increases. In another embodiment, the look-ahead distance is a function of vehicle speed and vehicle yaw rate. That is to say, when the speed of the vehicle increases, the distance ahead also increases; and when the vehicle's deflection speed increases, the distance ahead decreases accordingly. When the standardization error is relatively large, the look-ahead distance decreases, and when the standardization error is relatively small, the look-ahead distance increases. Similarly, when the distance difference between the target point and the advancing position is relatively large (or small), the look-ahead distance may decrease (or increase) correspondingly. In addition, the driving action can also be combined with the viewing distance. For example, the look-ahead distance for left/right turns can be reduced. When changing lanes, the look-ahead distance can be increased (or shortened) according to the preferred distance (or time) for completing the lane change. Finally, if the driving action performed by the vehicle is lane keeping, the look-ahead distance can be shortened when the vehicle is about to cross the lane line. When the vehicle is driving very close to the center of the lane, the look-ahead distance can be increased.

In one embodiment, in order to obtain the position of the target point, the steering control method firstly locates the target point according to the running action of the vehicle. When the vehicle keeps moving in the lane, the steering control method uses the central line of the driving lane as the criterion, and performs offset (including zero offset) positioning for the target point. When the vehicle is changing lanes, the steering control method uses the central line of the changed lane as the criterion, and performs offset (including zero offset) positioning on the target point. When the vehicle is turning left or right, the steering control method uses the central line of the lane after turning as the criterion, and performs offset (including zero offset) positioning on the target point. When the vehicle is dodging an obstacle in the driving lane, the target point is located in an available adjacent lane to the left or right. Finally, the steering control method calculates the position of the target point according to the center line of the lane, the offset position and the look-ahead distance.

In one embodiment, it is assumed that the vehicle is driving a distance ahead, and maintains the current vehicle speed and yaw speed. In this case, the steering control method predicts the forward position of the vehicle, and the current yaw rate depends on the current yaw rate of the vehicle and/or the current steering angle of the vehicle. Additionally, there is another way of predicting the forward position of the vehicle. When the vehicle is traveling the distance ahead, it is assumed that the vehicle maintains the current speed and steering angle. According to the above two hypothetical situations, the prediction of the forward position of the vehicle can be achieved by geometric relationship, kinematic model or mathematical model of vehicle dynamics (such as 2D vehicle model).

Compared with the prior art, the steering control method disclosed in the present invention has two main advantages. First, the disclosed steering control method uses the target point (or target line) as the basis of the control technique. In the condition of lane keeping, the target point (or target line) is positioned with an offset (including zero offset) based on the center line of the lane where the vehicle is driving. In other driving states, the target point (or target line) is positioned with an offset (including zero offset) based on the center line of the transformed lane. Therefore, for different driving actions and driving situations, the method disclosed in the present invention eliminates the need of planning or generating different ideal paths. Second, in the steering control method disclosed in the present invention, error normalization is first performed on the error between the predicted position and the target point. Then, the steering control method integrates the normalized error and multiplies the integrated value by the feedback gain. Error normalization increases the look-ahead distance of the disclosed steering control as vehicle speed increases without negatively affecting system stability. At the same time, the disclosed steering control can also shorten the look-ahead distance without reducing the feedback gain, so as to provide a suitable steering action when the vehicle is driving a sharp curve. Due to the above two advantages, the steering control method disclosed in the present invention provides a simple vehicle steering mechanism, and can ensure high precision and stability under different driving actions and driving conditions.

According to the steering control method described above, another embodiment of the present invention provides an improved control method. This method is not to obtain the target point at a certain distance in front of the vehicle, but to obtain the target line in front of the vehicle according to the driving action of the vehicle. Then, the method estimates the forward position of the vehicle at the look-ahead distance in front of the vehicle, and then calculates the distance difference between the forward position of the vehicle and the target line. Next, the distance difference between the forward position of the vehicle and the target line can be normalized to calculate the normalized error. Then, the normalized error is integrated, and then multiplied by the gain to obtain the steering control command. Accordingly, according to the generated steering control command, the actuator correspondingly steers the tires of the vehicle to drive the vehicle to perform a driving motion.

In one embodiment, the disclosed steering control method obtains the target line by first positioning the target line according to the driving action. In the condition of lane keeping, the target line is based on the center line of the lane where the vehicle is driving, and the offset (including zero offset) is positioned. In the state of changing the lane, the target line is based on the central line of the new lane after changing, and the offset (including zero offset) is positioned. When the vehicle is about to turn left or right, the target line is positioned with an offset (including zero offset) based on the central line of the new lane after turning. "Offset" can be positional and/or angular offset. Therefore, the method disclosed in the present invention calculates the position of the target line according to the corresponding central line and offset.

According to the steering control method disclosed in the present invention, the present invention further discloses a lateral control system. This system is installed on vehicles with steerable tires to control the steering of the vehicle. The lateral control system includes: a road detection device to provide road data ahead of the vehicle; a speed sensor to provide a vehicle speed signal; a steering angle sensor to provide a steering angle signal; a processor to calculate a steering angle command; and at least one steering The actuator, the actuator turns the tires according to the steering angle command, and then drives the vehicle to perform the desired driving action. The processor is connected to the road detection device to receive road data. The processor is also connected with the speed sensor to receive the vehicle speed signal. In addition, the processor is also connected to the steering angle sensor to obtain the steering angle signal. The steering actuator is connected to the steering angle sensor to receive the steering angle signal. The steering actuator is also connected to the processor to receive steering angle commands.

In one embodiment, the road detection device includes an image sensor and an image processing unit. The image sensor takes pictures of the road in front of the vehicle. With the image provided by the image sensor, the image processing unit calculates the shape of the road as road data. In another embodiment, the road detection device includes a satellite navigation system, a digital map and a processing unit. The satellite navigation system is responsible for determining the vehicle's location. The processing unit builds the vehicle's position into a digital map and provides data on the road ahead of the vehicle. In yet another embodiment, the road detection device includes a laser scanner and a processing unit. The laser scanner is responsible for transmitting laser light pulses and collecting reflections from objects in front of the vehicle. The processing unit determines the shape of the road through the collected reflected light as road data.

The processor determines the steering angle command according to the following procedures: determining the look-ahead distance; calculating the target point position of the look-ahead distance in front of the vehicle; estimating the forward position of the vehicle; calculating the distance difference between the position of the target point and the forward position of the vehicle; The distance difference is divided by the look-ahead distance to calculate the normalized error; and the normalized error is integrated.

In one embodiment, the processor determines the look-ahead distance according to at least one of the following: vehicle speed, vehicle yaw rate, vehicle lateral acceleration, vehicle steering angle, lane curvature, distance from the vehicle to the lane line, distance from the vehicle to the obstacle, normalized The error, the distance difference between the position of the target point and the forward position of the vehicle, and the driving action performed by the vehicle. Next, the processor determines the position of the target point through the following procedures: according to the road data provided by the road detection device, the central line of the road is estimated; Location. The target point is located at an offset position (including zero offset) based on the center line of the road, and within the look-ahead distance in front of the vehicle.

The way for the processor to estimate the forward position of the vehicle is: to estimate the vehicle's deflection speed according to the steering angle signal; In another embodiment, the lateral control system further includes a yaw speed sensor and a processor. The yaw speed sensor provides the vehicle's yaw speed signal. The processor estimates the forward position of the vehicle according to the vehicle speed signal provided by the speed sensor and the yaw speed signal provided by the yaw speed sensor.

In another embodiment, the lateral control system is connected to a driving maneuver decision unit mounted on the vehicle. The processor receives driving action commands (such as lane keeping, changing lanes, turning left, turning right, and avoiding obstacles) from the driving decision-making unit. The processor then selects the target line according to the driving motion command. The target point is located on the target line. The processor then calculates the position of the target point according to the target line and the look-ahead distance. According to the command of the driving action, the target line is positioned at the offset (including zero offset) position based on the center line of the lane the vehicle is driving on or the center line of the lane to be changed.

The steering control system disclosed in the present invention inherits the advantages of the steering control method disclosed in the present invention. Under different driving environments, these control systems can manipulate the vehicle to complete different driving actions with high accuracy and stability.

Description of drawings

FIG. 1 is a schematic diagram of an existing steering control method, which plans an ideal path for changing lanes;

Fig. 2 is a schematic diagram of a target point according to an embodiment of the present invention, which is the data on which the steering control is based, and then executes the driving action of changing lanes;

FIG. 3 is a schematic diagram of an existing steering control method, wherein a steering control command is determined according to a reference point along an ideal path;

Fig. 4 is a schematic diagram of steering control according to an embodiment of the present invention, which uses target points along the lane the vehicle is traveling on or changing to the lane;

FIG. 5 is a flowchart of an embodiment of the present invention, which includes steering control established according to a target point;

Fig. 6 is a schematic diagram of steering control according to another embodiment of the present invention, which uses the vehicle's driving lane or the shifted target line as a reference;

FIG. 7 is a flowchart of an embodiment of the present invention, which includes steering control established according to a target line;

8 is a schematic block diagram of a steering control system in a vehicle according to an embodiment of the present invention, wherein the steering control system automatically makes the vehicle follow the road it is traveling on;

9 is a schematic block diagram of a steering control system in a vehicle according to another embodiment of the present invention, wherein the steering control system automatically makes the vehicle follow the road it is traveling on;

10 is a schematic block diagram of a steering control system in a vehicle according to the present invention, wherein the steering control system automatically makes the vehicle perform different driving actions.

Among them, reference signs:

102, 202 vehicles

Lanes 104 and 106

108, 110 Central Line

112 ideal path

114, 116, 118 Reference points

204, 206 target points

208 actual route

602 target line

A1, A2, A3, L1, L2, L3 positions

d front view distance

B2, P2 vehicle forward position

R2 reference point

T2 target point

e, ε error

500, 700 programs

Step 502 Obtain the position of the target point

Step 504, Step 704 Predict the forward position

Step 506 Calculate the distance between the target point and the advancing position

Step 508, Step 708 Calculate steering control command

Step 702 Obtain the position of the target line

Step 706 Calculate the distance from the forward position to the target line

800, 900, 1000 lateral control system

802 Road detection device

804 speed sensor

806 Steering angle sensor

808 Lateral Control Processor

810 Estimated Target Position Module

812 Determining the look-ahead distance module

814 Estimated forward position module

816 Decide to turn to the instruction module

1002 Select target line module

818 steering actuator

902 Yaw speed sensor

904, 1004 Lateral Control Processor

specific embodiment

The invention discloses a vehicle steering control method and system. The method and system can perform various driving actions without planning an ideal path. The method and system use the target located on the road the vehicle is traveling on, the road to change or the road to turn into as a reference point for control. In order to help explain the difference between the present invention and the existing method, FIG. 1 is a schematic diagram of the method adopted in the prior art, which plans an ideal path for changing lanes. FIG. 2 is a schematic diagram of a target point according to an embodiment of the present invention, which is used as a basis for steering control, and then executes a driving action of changing lanes.

In FIG. 1 , the vehicle 102 uses an existing automatic steering control system (not shown). Before reaching position A1 , vehicle 102 travels along lane centerline 108 of lane 104 . Existing systems use the lane center line 108 as an ideal path, and determine the steering control command through a plurality of reference points 114 along the ideal path (lane center line 108 ). At position A1 , vehicle 102 needs to change into lane 106 . Thus, the ideal path 112 is planned or generated by existing steering control systems. With the ideal path 112 , the vehicle 102 can smoothly change to the lane 106 . Then, the existing steering control system uses the reference point 116 located on the ideal path 112 to determine the steering control command and guide the vehicle 102 to travel on the ideal path 112 . When the vehicle 102 completes the lane change at position A3 , the existing steering control system changes the ideal path to the centerline 110 of the lane 106 . Thus, for subsequent lane keeping, existing steering control systems locate a reference point 118 along the centerline 110 of the ideal path.

In FIG. 2 , a vehicle 202 uses an automatic steering control system (not shown) according to an embodiment of the present invention. Similar to the situation in FIG. 1 , the vehicle 202 first follows the lane 104 before reaching the location L1 . Vehicle 202 then changes to lane 106 and continues to follow lane 106 . However, unlike existing steering control systems, the automatic control system disclosed in the present invention does not plan an ideal path for lane keeping or lane changing; on the contrary, the automatic control system of the present invention directly uses multiple target points as steering control , and these target points are set along the lane where the vehicle 202 is traveling or the lane to be changed. Before arriving at the position L1 , the driving behavior performed by the vehicle 202 is lane keeping. Therefore, a plurality of target points 204 are established along the lane 104 , which is the road on which the vehicle 202 travels. In one embodiment, the target point 204 is established along the lane centerline 108 . In another embodiment, the target point 204 is set at a position offset from the lane center line 108 . When the vehicle 202 is at the position L1 and after it is at the position L1 , the driving action of the vehicle 202 is to change lanes, and at this time, the positions of the plurality of target points 206 are established along the lane 106 . The lane 106 is the lane to which the vehicle 202 intends to switch. In one embodiment, the location of the target point 206 is established along the lane centerline 110 . In another embodiment, the target point 206 is set at a position offset from the lane center line 110 . By using the target point as a reference for steering control, the automatic steering control system disclosed in the present invention can smoothly guide the vehicle 202 to change from the lane 104 to the lane 106 . That is to say, the actual driving route 208 of the vehicle 202 is an after effect of the steering control, rather than a planned route before changing lanes. After completing the lane change, the vehicle 202 follows the lane 106 while the target point 206 remains on the lane 106 .

The difference between the present invention and the prior art for determining the steering control command can be clearly illustrated by FIG. 3 and FIG. 4 . FIG. 3 is a schematic diagram of a steering control method in the prior art, wherein multiple reference points along an ideal path are used. In FIG. 3 , vehicle 102 is at position A2 and is changing lanes. As shown in FIG. 1 , the conventional steering control method uses a reference point 116 along the ideal path 112 as a control reference. At each time occasion t, prior art steering control methods use a specific reference point along the ideal path and at a predetermined distance in front of the vehicle. The predetermined distance can be represented by the look-ahead distance d. Therefore, when the vehicle 102 is at the position A2, the specific reference point 116 for determining the steering control command is R2. Prior art steering control methods then predict the forward position of the vehicle 102 using either the primary prediction or the secondary prediction. For example, FIG. 3 exemplifies the case of the secondary prediction, so the forward position of the vehicle is B2 and is set in front of the vehicle 102 . Wherein the distance between the vehicle 102 and the forward position B2 of the vehicle is the aforementioned front-view distance d. Next, the prior art steering control method calculates e(t), which is the distance between the reference point R2 and the vehicle forward position B2. Therefore, the calculation formula of the steering control command is δ(t)=k·(Σe(t)). It should be noted that the vehicle 102 is not centered on the position A2 of the ideal path, and the direction angle of the vehicle 102 is not facing the tangent direction of the ideal path. This is because the actual path or trajectory of the vehicle usually deviates from the ideal path due to residual errors.

FIG. 4 is a schematic diagram of a steering control method according to an embodiment of the present invention. In this figure, vehicle 202 is at location L2 and is changing lanes. As shown in FIG. 2 , the steering control method uses a plurality of target points 206 as references for control. These target points 206 are located in the lane that the vehicle is driving (when the driving action of the vehicle 202 is lane keeping) or the lane to be changed (when the driving action of the vehicle 202 is lane changing). Since the vehicle 202 is changing lanes, the location of the target point 206 is set in the lane 106 , ie, along or offset from the lane centerline 110 of the lane 106 . At each time occasion t, the steering control method uses a specific target point, which is set at a look-ahead distance d(t) in front of the vehicle. (An advantage of the method disclosed in the present invention is that the look-ahead distance d is variable, and the change of the look-ahead distance d will not cause system instability or sacrifice performance. This advantage will be discussed later.) Therefore, When the vehicle is at position L2, the specific reference point 206 for determining the steering command is T2. In one embodiment, the next step of the steering control method is to estimate the forward position of the vehicle assuming the vehicle maintains the current vehicle speed and yaw rate (yaw rate). Therefore, the forward position of the vehicle is P2. The vehicle forward position P2 is set in front of the vehicle 202 , wherein the distance between the vehicle 202 and the vehicle forward position P2 is the aforementioned front-view distance d(t). Next, the steering control method calculates an error value ε(t). The error value ε(t) is the distance difference between the target point T2 and the advancing position P2. Then, the steering control method disclosed in the present invention performs normalization on the error value ε(t). The way of standardization is to divide the error value ε(t) by the viewing distance d(t), and then integrate the standardized error ε(t): δ(t)=k(t)·(Σ(ε(t)/ d(t))).

By using the look-ahead distance d(t) to normalize the error ε(t), the steering control method disclosed in the present invention has considerable advantages compared with the prior art. As mentioned earlier, when the vehicle is driving on a sharp curve, the steering control system operating based on the main prediction cannot achieve satisfactory performance. The main prediction assumes that the vehicle is traveling in a straight line. That is to say, when the ideal path includes sharp curves, the above assumptions are quite different from the actual situation. When using secondary predictions, prior art steering control methods become unstable at high vehicle speeds and fail to keep the vehicle on the desired path. (From the point of view of control theory, a control system needs to have sufficient phase margin and gain margin to be stable. In order to achieve sufficient phase margin, the control system needs to have sufficient phase lead. In order to achieve sufficient gain margin, the feedback gain of the control system should not exceed a specific value determined by the gain margin.) At high vehicle speeds, the look-ahead distance suitable for low vehicle speeds is not enough to provide sufficient phase lead for the steering control system. In the absence of sufficient phase lead, the steering control system tends to become unstable. A possible remedy is to increase the look-ahead distance as the vehicle speed increases. However, under the same ideal path and predicted path conditions, the longer the look-ahead distance, the larger the error ε. Therefore, increasing the viewing distance time is essentially equal to increasing the feedback gain. The above situation also causes the steering control system to lose the gain margin, which in turn causes the system to be unstable.

The steering control method disclosed in the present invention assumes that the vehicle maintains the current yaw speed (or steering angle), and then predicts the forward position of the vehicle. Therefore, when the vehicle is running on a straight lane or a sharp curve, the steering control method disclosed in the present invention can predict the forward position of the vehicle. More importantly, the error value ε is normalized by the look-ahead distance, and the result is integrated to calculate the steering control command. At higher vehicle speeds, the steering control method disclosed in the present invention can increase the look-ahead distance d, thereby introducing more phase lead into the system. Although the increase of the look-ahead distance will also increase the error value, the normalization of the error value (error divided by the look-ahead distance) will remove the negative effect of increasing the look-ahead distance in the prior art. In short, by standardizing the error value, the steering control method disclosed in the present invention can flexibly adjust the look-ahead distance d(t), thereby maintaining system stability and performance in a wide range of vehicle speeds.

As mentioned earlier, another disadvantage of prior steering control based on secondary prediction is that the vehicle cannot maintain the desired path in tight cornering situations. Applicable look-ahead distances on straight lanes or gentle curves may be too long for sharply changing paths. As a result, the reference point would be too far away, and the actual curves of the ideal path would not be captured. One possible remedy is to reduce the length of the look-ahead distance when the vehicle is driving on a sharp curve. However, in the case of the same ideal path and estimated path, the error value ε is smaller when the path ahead is shorter. Therefore, shortening the viewing distance is essentially equivalent to reducing the feedback gain. The above situation causes the control system of the prior art to fail to provide appropriate steering control commands when the vehicle needs to be driven on a sharply curved lane.

The limitations of prior art steering control methods are also overcome by error normalization in the steering control method disclosed in the present invention. When driving on a sharp curve, the steering control system disclosed by the present invention can shorten the distance ahead. Although the shortening of the look-ahead distance also leads to a smaller error value ε, it has little effect on the feedback gain, mainly because the smaller error value ε is normalized by the smaller look-ahead distance. Therefore, the steering control system disclosed in the present invention has enough feedback gain to provide sufficient steering control when the vehicle is driving a sharp curve.

To sum up, compared with the prior art, the steering control method disclosed in the present invention has two advantages. First, the steering control method disclosed in the present invention uses the target point as the reference of control, and the target point is located on the lane the vehicle is driving (when the driving action is lane keeping) or the lane to be changed (when the driving action is other driving actions) ). Therefore, for different driving actions and driving conditions, the steering control method disclosed in the present invention does not need to plan or generate different ideal paths. Second, the steering control method disclosed in the present invention performs normalization on the error between the predicted position and the target point, integrates the normalized error, and then multiplies the integral of the normalized error by the feedback gain to calculate the steering control command. Through error normalization, when the vehicle speed increases, the steering control disclosed in the present invention allows the look-ahead distance to increase accordingly, but without sacrificing system stability. In addition, the error standardization also enables the steering control method disclosed in the present invention to reduce the look-ahead distance without reducing the feedback gain. Therefore, the vehicle has sufficient steering control when driving on sharp curves. Analysis based on control theory shows that, regardless of vehicle speed (and road curvature), two open-loop zeroes are introduced with an ideal and fixed damping ratio due to error normalization. Since the poles of the closed-loop control system approach the open-loop zero, error normalization allows the closed-loop control system to maintain high feedback gain over a wide range of vehicle speeds without sacrificing system stability. In short, under different driving actions and driving situations, the steering control method disclosed in the present invention provides a simple vehicle steering mechanism with high precision and stability.

FIG. 5 is a flowchart of a procedure 500 of a steering control method according to an embodiment of the present invention. The program 500 is built into a processor installed in the vehicle. The processor adopts the real-time operation method and uses a preset processing cycle, such as 10ms. The process 500 begins with step 502, which is to obtain the position of the target point. According to the ongoing driving action of the vehicle, the process 500 first determines the lane where the target point is located. When the driving action performed by the vehicle is lane keeping, the target point is located within the lane the vehicle is driving. When the vehicle is changing lanes, the target point is located in the lane to be changed (as shown in FIG. 2 and FIG. 4 ). When the vehicle is turning left or right, the target point is located in the lane the vehicle is turning into. When the vehicle is performing obstacle avoidance to avoid obstacles in the current lane, the target point is located in the available lane on the left or right. In one embodiment, the target point is always located at the center line of the lane. In another embodiment, the target point is located at a position offset from the center line of the lane, where the offset is not fixed and can be zero offset or non-zero offset.

When the program 500 has determined the lane where the target point is located, the program 500 then determines the look-ahead distance, and uses at least one of the following factors as the basis for determining the look-ahead distance: vehicle speed, vehicle yaw speed, vehicle lateral acceleration, vehicle steering angle, Lane curvature, the distance between the vehicle and the lane edge, the distance between the vehicle and the obstacle, the normalization error, the distance difference between the target point and the forward position, the driving action of the vehicle and the position of the obstacle (if the driving action is an obstacle object dodges). In one embodiment, the look-ahead distance is a function of vehicle speed. For example, d(t)=a·v(t), where d is the look-ahead distance, v is the vehicle speed, and a can be a constant or a variable gain. In another embodiment, the lower limit of the look-ahead distance can also be expressed by the following formula: d(t)=max(a·v(t), dmin). That is to say, if a·v(t)>dmin, then d(t)=a·v(t); otherwise d(t)=dmin. In yet another embodiment, the front distance d(t) is a function of the speed of the vehicle and the yaw speed of the vehicle: d(t)=f(v(t), ω(t)), where ω(t) is the yaw speed of the vehicle (yaw rate). The function f(v(t), ω(t)) is designed so that when the vehicle speed increases, the distance ahead d(t) also increases; and when the vehicle yaw speed increases, the distance d(t) ahead decreases. In another embodiment, the steering angle is used instead of the yaw speed to determine the look-ahead distance d(t). In yet another embodiment, the look-ahead distance d(t) is determined according to the vehicle speed and the curvature of the lane. The greater the curvature, the shorter the viewing distance d(t). In addition, when the standardized error becomes larger, the look-ahead distance also becomes shorter. When the standardized error ratio becomes smaller, the look-ahead distance becomes longer. Similarly, when the distance difference between the target point position and the advancing position is relatively large (or small), the look-ahead distance may decrease (or increase) correspondingly. (Because the look-ahead distance of each processing cycle is determined before calculating the normalized error or the distance difference between the target point and the forward position. Therefore, the normalized error of the previous cycle or the distance between the target point and the forward position The difference in distance can be used to determine the look-ahead distance of the next cycle.) Finally, if the vehicle’s ongoing driving action is lane keeping, when the vehicle is about to cross the lane edge (relatively far from the center of the lane), the look-ahead distance can be reduced . The look-ahead distance can be increased when the vehicle is driving closer to the center of the lane. Also, vehicle driving actions are also incorporated in the following way: For example, for left and right turns, the look-ahead distance d(t) can be reduced. For lane changes, the look-ahead distance can be increased if a longer distance (or more time) is the priority for changing lanes. Conversely, if the lane change is to be done at a shorter distance, the look-ahead distance can be reduced.

After the lane-to-view distance d(t) has been determined, the procedure 500 then calculates the position of the target point located on the centerline (or offset from the centerline) of the lane. The distance from the target point to the vehicle is equal to the front-view distance d(t). In one embodiment, the position of the target point is obtained by solving two formulas: the first formula represents the lane centerline (or offset centerline), and the second formula represents the distance between the target point and the vehicle, It is equal to the look-ahead distance.

Next, in step 504, the routine 500 estimates the vehicle's forward position. The forward position is also at a look-ahead distance d(t) in front of the vehicle. The estimated forward position is based on the assumption that the vehicle will maintain its current speed and yaw velocity when the vehicle travels the look-ahead distance. In other words, the vehicle is traveling on an arc-shaped path, and its radius is R=v(t)/ω(t), where v(t) is the vehicle speed, ω(t) is the vehicle yaw speed or yaw rate, and the arc length It is the viewing distance d(t). The end of the arc is the estimated forward position. Therefore, based on a simple geometric relationship, the location of the advancing position can be obtained. In one embodiment, the yaw rate of the vehicle is directly used in the estimation process. In another embodiment, the steering angle is used as a tool for estimating the yaw speed.

In step 506, the distance between the target point and the advancing position is calculated as an error value ε(t). Next, in step 508, the program 500 determines the steering control command in the following manner: (1) Use the look-ahead distance d(t) to normalize the error value ε(t), and then obtain the normalized error ε(t)/d(t ), (2) the normalized error integral: Σ(ε(t)/d(t)), (3) multiply the normalized error integral by the gain, and then generate the steering control instruction: δ(t)=k(t)·( Σ(ε(t)/d(t))). The actuator installed on the vehicle receives the generated steering control command, and operates the tires according to the steering control command to make the vehicle perform the required driving action.

In another embodiment of the lateral control method, the target point is replaced by a target line. Then, the distance from the estimated forward position to the target line is calculated as an error value ε(t), which is used in the steering control command: δ(t)=k(t)·(Σ(ε(t)/d(t ))). In this embodiment, the target line in front of the vehicle is obtained according to the driving action performed by the vehicle. When the vehicle is performing a lane keeping action, the target line is located at the offset (including zero offset) lane center line of the lane the vehicle is driving on. (When the offset is zero, the target line is the center line of the lane the vehicle is traveling on.) When the vehicle is changing lanes, the target line is located at the offset (including zero offset) center line of the lane the vehicle wants to change . When the vehicle is about to turn left or right, the target line is located at the offset (including zero offset) central line of the lane into which the vehicle intends to turn. When the vehicle is avoiding obstacles, the target line is located at a position offset from the center line of the lane in which the vehicle is running. The offset position of the target line depends on the size and location of the obstacle and the availability of the left and right lanes of the lane in which the adjacent vehicle is traveling. Offset itself can be classified into position offset and/or angle offset. When the offset is a position offset, the target line and the lane centerline are parallel to each other, and there is a distance (including zero distance) between them. When the offset is an angle offset, the target line is positioned by rotating the centerline of the lane to an angle. When the offset includes position offset and angle offset at the same time, the positioning of the target line is through: first, the center line of the lane is shifted parallel to the position offset, and then rotated to a specific angle offset.

FIG. 6 is a schematic diagram of an embodiment of a steering control method disclosed in the present invention. Vehicle 202 in this figure is changing from lane 104 to lane 106 . Vehicle 202 in the lane change is at position L2. Accordingly, the target line is positioned at a position offset (including zero offset) from the lane centerline 110 of the lane 106 . The lane 106 is also the road on which the vehicle 202 will change. In the embodiment shown in the figure, the offset is zero position offset, and the target line 602 is set at the central line 110 of the lane 106 . The steering control method of the present invention then estimates the forward position P2 of the vehicle, which is located at the distance d ahead, as shown in FIG. 4 . The look-ahead distance d depends on at least one of the following items: vehicle speed, vehicle yaw velocity, vehicle lateral acceleration, vehicle steering angle, lane curvature, normalization error, distance difference from the forward position to the target line, vehicle driving action and the size of the obstacle and position (if the driving action is obstacle avoidance). Then, the steering control method of the present invention finds the distance difference between the advance position P2 and the target line 602, and is called an error term ε(t). Next, the steering control method of the present invention calculates a normalized error. It is calculated by dividing the distance difference by the viewing distance, that is, ε(t)/d(t). After that, the normalized error is integrated and multiplied by the gain to calculate the steering control command: δ(t)=k(t)·Σ(ε(t)/d(t)). Finally, the actuator installed on the vehicle receives the generated steering control command, and manipulates the tires according to the steering control command to make the vehicle perform the required driving action.

FIG. 7 is a flowchart showing a procedure 700 of a steering control method according to an embodiment of the present invention, and the procedure 700 is based on a target line, the procedure 700 . Procedure 700 is different from procedure 500 in FIG. 5 . Specifically, the procedure 700 obtains the position of the target line (instead of the target point) according to the vehicle driving action in step 702 . The sequence for program 700 to obtain the position of the target line is: (1) locate the target line according to the driving action of the vehicle, and (2) calculate The location of the target line. Then, in step 704, the process 700 determines the look-ahead distance and estimates the forward position of the vehicle. The vehicle forward position is set within a look-ahead distance in front of the vehicle. In step 706, the process 700 calculates the distance from the advance position to the target line. Finally, in step 708, the process 700 calculates the steering control command. The calculation method is: (1) normalize the error term (ε(t)/d(t)), (2) integrate the normalized error value, (3) multiply the integral of the error value and the gain (δ(t )=k(t)·Σ(ε(t)/d(t))).

FIG. 8 is a schematic block diagram of a lateral control system 800 installed in a vehicle. The vehicle itself has multiple tires which control the direction the vehicle is going. The lateral control system 800 includes a road detection device 802 , a speed sensor 804 , a steering angle sensor 806 , a lateral control processor 808 and at least one steering actuator 818 . The road detection device 802 provides road data ahead of the vehicle. The speed sensor 804 provides a vehicle speed signal. Steering angle sensor 806 provides a steering angle signal. The lateral control processor 808 determines the steering angle command. At least one actuator 818 is responsible for deflecting the tires and causing the vehicle to perform the desired driving maneuver according to the generated steering control commands.

In one embodiment, the road detection device 802 includes an image sensor and an image processing unit. The image sensor collects images in front of the vehicle, and the image processing unit calculates the shape of the road based on the collected image data. For example, roads may be defined by lane markings, where the lane markings make up markings (straight or curved). These markings (straight or curved) can be represented by mathematical equations, such as polynomial equations. Therefore, in one embodiment, the image processing unit recognizes the lane markings from the image data collected by the image sensor, and determines the parameters of the polynomial equation according to the lane markings. Therefore, the road data provided by the image detection device 802 includes polynomial parameters of the lane markings. The lane markings here generally refer to the left and right markings of the lane on which the vehicle is traveling.

In another embodiment, the road detection device 802 includes a satellite navigation system, a digital map and a processing unit. The satellite navigation system determines the vehicle's location. The types of satellite navigation systems include GPS/Global Positioning System (Global Positioning System), GLONASS/Global Navigation Satellite System (Global Navigation Satellite System), Galileo Positioning System (Galileo Positioning System) and Beidou Navigation Satellite System/BDS (Beidou Navigation Satellite System). Digital maps contain road data. The processing unit establishes the vehicle position in a digital map and provides road data ahead of the vehicle position. The satellite navigation system provides the position of the vehicle in a geographic coordinate system such as a longitude, latitude and altitude coordinate system. By establishing the location of the vehicle in the digital map, the processing unit can identify the location of the vehicle in the map and obtain surrounding road data. Road data may include road direction, road curvature, lane location, number of lanes, lane width, node locations, distance between road intersections (junctions), and the like.

In a third embodiment, the road detection device 802 includes a laser scanner and a processing unit to provide road data. A laser scanner sends laser pulses and captures the beams reflected from objects in front of the vehicle. The processing unit determines the distance between the vehicle and the object according to the principle of time-of-flight ranging (Time-Of-Flight/TOF). The measurement rate (reflected light captured) of lane markings is high compared to full ground measurements due to the high reflectivity of lane markings. Therefore, the processing unit can detect the lane markings according to the rate of beam reflection. Correspondingly, lane markings (lines or curves) can be represented via different equations, such as polynomial equations. Next, the processing unit calculates the parameters of the polynomial equation based on the detected lane markings. Therefore, the road data includes the polynomial parameters of the lane markings, which are usually the lane markings on the left and right sides of the lane in which the vehicle is moving.

The lateral control processor 808 is connected to the road detection device 802 to receive road data. The lateral control processor 808 is also connected to the speed sensor 804 to receive the vehicle speed signal. In addition, the lateral control processor 808 is also connected to the steering angle sensor 806 to receive the steering angle signal. Depending on the position of the steering angle sensor, the steering angle signal is the travel angle of the tires, the angle of the steering wheel or the angle of the axis of rotation between the steering wheel and the tires in the steering system.

To determine the steering angle command, lateral control processor 808 uses a procedure similar to procedure 500 in FIG. 5 . The lateral control processor 808 first determines the look-ahead distance d in block 812 . As described in step 502 of Figure 5, the look-ahead distance is determined based on at least one of the following items: vehicle speed, vehicle yaw rate (vehicle yaw rate), vehicle lateral acceleration, vehicle steering angle, lane curvature, normalized error, forward position The distance difference from the target point, the distance from the vehicle to the lane edge (when the driving behavior is lane keeping), the vehicle driving behavior, and the position of the obstacle (when the driving behavior is obstacle avoidance).

In block 810, the lateral control processor 808 estimates the position of the target point. The position of the target point is set at a position offset (including zero offset) from the center line of the lane, and also within the look-ahead distance in front of the vehicle. In one embodiment, the lateral control system 800 only performs lane keeping maneuvers. Therefore, the lane center line is always the center line of the lane on which the vehicle is running. (In other embodiments, the lateral control system 800 performs other driving maneuvers, including lane keeping, lane changing, left/right turning, and obstacle avoidance. The details of this embodiment will be discussed later with FIG. 10 .)

In one embodiment, the lateral control processor 808 first estimates the center line of the lane according to the road data provided by the road detection device 802 , and then further estimates the position of the target point. As mentioned earlier, if a video or laser scanning type road detection device is used, the road data will include the road shape, ie an equation representing the left and/or right lane markings. Therefore, the center line of the lane can be represented by the same type of equation (eg polynomial equation). Furthermore, from the equations for the left and/or right lane markings, the parameters of the equations representing the lane centerlines can be estimated. If the road detection device uses satellite navigation technology, the road data will include road direction, curvature, lane position, number of lanes, lane width, etc. The centerline of a lane can be estimated as an arc whose radius is 1 divided by the road curvature, and whose tangent direction is the road direction. The starting point of the arc is the center of the lane in which the vehicle is traveling. With any of the above road detection devices, the lane centerline is estimated as an equation representing the centerline (straight line or curve).

Next, the lateral control processor 808 calculates the position of the target point according to the centerline of the lane and the look-ahead distance. Basically, the calculation process includes the solution of the equation for the center line of the lane (for example: f(x,y)=0) and the solution of the equation for the distance ahead (x ² +y ² =d ² ). The solution (Tx, Ty) located in front of the vehicle (Tx>0) is the position of the target point. In this embodiment, the target point is located within the centerline of the lane and within the look-ahead distance in front of the vehicle. In another embodiment, the target point may be located offset from the lane centerline and within a look-ahead distance in front of the vehicle. In the case of this embodiment, the calculation process includes finding the solution of the equation for offsetting the central line (for example: f (x, y, m) = 0, where m is the offset amount) and the solution of the equation representing the viewing distance Solve (x ² +y ² =d ² ).

In another embodiment, the lateral control processor 808 directly estimates the offset of the lane center line according to road data without estimating the lane center line. For example, road data includes equations for left and/or right lane markings. The lateral control processor 808 then calculates the location of the target point by solving the equation for the offset from the center line of the lane and the equation for the distance ahead.

In block 814, the lateral control processor 808 predicts the forward position of the vehicle. The prediction process is based on the steering angle signal and the look-ahead distance. In one embodiment, the assumption used in the estimation is that the vehicle maintains the current vehicle speed and yaw rate (yaw rate). That is, the radius of the vehicle is constant as it travels along the arc. According to the geometric relationship of the steering angle, the radius of the arc is estimated as: R=L/tan(δ)=L/tan(α·δ _meas ). In this formula, L is the wheelbase of the vehicle, δ is the steering angle of the tire, δ _meas is the steering angle data provided by the steering angle sensor, and α is the tire steering angle and steering angle data (provided by the steering angle sensor )proportion. The value of α is influenced by the installation position of the steering angle sensor (eg in the steering column or in the power steering unit). When the position of the steering angle sensor is determined, α can be known.

In another embodiment, the vehicle yaw velocity is estimated by the vehicle model based on the steering angle. An example of a vehicle model is the well-known 2D vehicle model. Under the framework of the two-dimensional vehicle model, the vehicle deflection velocity is the state of the model, and the vehicle speed is the parameter of the model. The model takes the steering angle as input and estimates the yaw velocity of the vehicle as one of the states of the model. On the other hand, when a vehicle dynamics model is used, time-series data of vehicle yaw speeds ω(t) to ω(t+T) can be estimated. The assumption for this estimation is that the vehicle maintains the current speed v(t) and steering angle δ(t) during the time period [t,t+T], where T can be set as d(t)/v(t) . That is, T is the time it takes for the vehicle to complete the travel distance d(t). Therefore, according to the vehicle speed and the estimated vehicle yaw velocity, the forward position of the vehicle can be determined by the following equation:

x(t _k+1 )＝x(t _k )+v(t)·(t _k+1 -t _k )·cos(θ(t _k ));

y(t _k+1 )=y(t _k )+v(t)·(t _k+1 -t _k )·sin(θ(t _k ));

θ(t _k+1 )=θ(t _k )+ω(t _k )·(t _k+1 −t _k ).

In the state where the coordinates are fixed to the vehicle, the initial state of the model is (x(t ₀ ), y(t ₀ ),θ(t ₀ ))=(0,0,0). Please note that in this embodiment, during the period [t, t+T], since the vehicle is assumed to maintain the current speed, the speed is always v(t). In another embodiment, during the period [t, t+T], the speed can be estimated according to the current acceleration a(t) or the acceleration curve. Therefore, the forward position of the vehicle can be estimated by the following equation:

x(t _k+1 )＝x(t _k )+v(t _k )·(t _k+1 -t _k )·cos(θ(t _k ));

y(t _k+1 )=y(t _k )+v(t _k )·(t _k+1 -t _k )·sin(θ(t _k ));

θ(t _k+1 )＝θ(t _k )+ω(t _k )·(t _k+1 -t _k );

v(t _k+1 )=v(t _k )+a(t _k )·(t _k+1 −t _k ).

In one embodiment, the starting point of the arc is the current position of the vehicle, and the tangent of the arc is the direction of the vehicle. When the estimated target point is the fixed coordinate of the vehicle, the current position of the vehicle at this coordinate is (0,0), and the tangent is on the x-axis. Therefore, arcs can be specifically defined by the function g(x,y)=0. Correspondingly, by solving the equations g(x,y)=0 and x ² +y ² =d ² , the forward position within the look-ahead distance can be determined. The obtained solution (Px, Py), where Px>0, is the advancing position.

After the target position (Tx, Ty), the forward position (Px, Py) and the look-ahead distance d have been determined, the lateral control processor 808 calculates the steering command at block 816 . The processor 808 first calculates the distance difference between the target point position and the advancing position: ε=sqrt((Tx−Px) ² +(Ty−Py) ² ). Next, the distance difference is divided by the look-ahead distance (ε/d), the processor 808 calculates the normalized error, and then integrates the normalized error value: Σ(ε/d). Finally, the processor 808 multiplies the integral of the normalized error by a certain gain (δ=k·Σ(ε/d)) to determine the steering angle command.

Since the distance difference ε and the look-ahead distance d are not fixed values, it is more appropriate to represent the steering command by δ(t)=k·Σ(ε(t)/d(t)). In another embodiment, the lateral control processor 808 further adjusts the gain k according to at least one of the following items: distance difference ε, look-ahead distance d, normalized error ε/d, vehicle speed, steering angle and road Camber. Therefore, the gain k is also a variable, so the steering command can be further rewritten as: δ(t)=k(t)·Σ(ε(t)/d(t)). In one embodiment, when the normalized error ε/d is greater than a certain critical value, or the normalized error has a large change, the lateral control processor 808 may decrease the gain k. In another embodiment, the lateral control processor 808 may increase the gain k when the steering angle is relatively large (eg greater than a certain threshold). Similarly, the lateral control processor 808 may increase the gain k when the road is relatively curved. (The lateral control processor 808 can obtain the road curvature through the equation of the center line of the lane. Another method is, if a satellite navigation type road detection device 802 is used, the lateral control processor 808 can obtain the road curvature provided by the road detection device 802 road data, directly obtain the road curvature.)

The lateral control processor 808 outputs the steering command to the steering actuator 818 after the calculation of the steering command is completed in block 816 . Steering actuator 818 steers the tires to keep the vehicle in the lane according to the steering command δ. In one embodiment, the steered tires are the front wheels of the vehicle. The steering actuator 818 steers the front wheels of the vehicle in accordance with the steering command δ. In another embodiment, steerable tires include front and rear wheels, and lateral control system 800 includes a pair of steering actuators. The front actuator steers the front wheels, while the rear actuator steers the rear wheels. The front actuator can make the steering angle of the front wheels to δ _f , and the rear actuator can make the steering angle of the rear wheels to δ _r , where the two angles satisfy the following relationship: δ _f −δ _r =δ. In one embodiment, the steering angles of δ _f and δ _r respectively satisfy the relationship of δ _r =(c/(Lc))·δ _f , where L is the wheelbase of the vehicle, and 0<c<L. Therefore, by finding the solutions of the above two equations (δ _f -δ _r = δ and δ _r = (c/(Lc))·δ _f ), the two actuators can determine the corresponding steering angle and compare The front and rear wheels are steered.

FIG. 9 is a schematic block diagram of a lateral control system 900 . The lateral control system 900 is installed on the vehicle. The vehicle includes a plurality of steering tires that control steering of the vehicle. Embodiments of lateral control system 900 differ from lateral control system 800 of FIG. 8 . The difference is that the lateral control system 900 uses a yaw rate sensor 902 (yaw rate sensor) to measure the yaw rate of the vehicle. The lateral control processor 904 then uses the yaw rate signal provided by the yaw rate sensor 902 to predict the forward position of the vehicle at block 814 . The prediction made by the lateral control processor 904 is based on the assumption that the vehicle maintains the current speed and yaw rate. That is, the radius of the vehicle does not change as it travels along the arc. Through the vehicle speed signal provided by the speed sensor 804 and the yaw speed signal provided by the yaw speed sensor 902, the radius R of the arc can be determined by the formula R=v/ω, where v is the vehicle speed and ω is the deflection speed. The starting point of the arc is the current position of the vehicle, and the tangent to the starting point of the arc is the forward direction of the vehicle. In one embodiment, vehicle-fixed coordinates are used. Therefore, the starting point of the arc is (0,0), and the tangent is the x-axis. Therefore, circular arcs can be defined individually via radii. In addition, the advance position (Px, Py) can be calculated by finding the solutions of the arc equation (g(x,y)=0) and the look-ahead distance equation (x ² +y ² =d ² ).

The lateral control system 800 in FIG. 8 and the lateral control system 900 in FIG. 9 execute the driving action of lane keeping, wherein, in terms of the lane on which the vehicle is driving, the target point is always located in the offset (including zero offset) lane The location of the central line. Please refer to FIG. 10 , which is a schematic block diagram of a steering control system 1000 installed in a vehicle. The steering control system 1000 automatically steers the vehicle to perform different driving maneuvers. Compared with the lateral control system 800 in FIG. 8 , the lateral control system 1000 is further connected to a unit (not shown in the figure) to receive a driving action command. Then, the lateral control processor 1004 determines a steering command to perform a driving maneuver. The unit that provides driving action instructions can be an autonomous decision-making unit, whose functions are to: receive the starting position and destination position (for example: provided by the driver), determine the path between the starting point and the destination, monitor the surrounding driving environment, and The driving action that the vehicle needs to perform is determined under the constraints of following the path to be driven and ensuring safety.

In order to perform specific driving maneuvers (eg, lane keeping, lane changing, turning left, turning right, and obstacle avoidance), the lateral control processor 1004 further includes a target line selection module 1002 . The target line selection module 1002 selects the target line according to the driving action command and the road data provided by the road detection device 802 . According to the driving action instruction, the target line is located at a position offset (including zero offset) from the center line of the lane in terms of the lane the vehicle is driving or the lane to be changed. For example, if the driving action command is lane keeping, the target line is located at a position offset (including zero offset) from the center line of the lane in terms of the lane the vehicle is driving on. If the driving action instruction is to change to the left (or right) lane, then for the left (or right) lane of the vehicle, the target line is located at a position offset (including zero offset) from the center line of the lane. If the driving action command is to turn left (or turn right), then in terms of the lane to be changed by the vehicle, the target line is located at a position offset (including zero offset) from the center line of the lane. If the driving action instruction is to use the left (or right) lane for obstacle avoidance, then for the left (or right) lane, the target line is located at the offset (including zero offset) lane center line . Just as the center line of the lane can be a straight line or a curve, the target line can also be a straight line or a curve.

In one embodiment, the offset is a positional offset, so the target line is parallel to the center line of the lane. When the position offset is zero, the target line is also the center line. In another embodiment, the lateral control processor 1004 makes progressive corrections to the offset according to driving situations and conditions. For example, lateral control processor 1004 may increase the offset when the vehicle enters a curve. If the vehicle is traveling in a curve, the offset remains unchanged. Lateral control processor 1004 may reduce the offset as the vehicle exits the curve. Through the above-mentioned adjustable offset, the vehicle can adopt the method of "cutting the curve and straightening" when cornering, so that the vehicle can drive on a relatively gentle curve route.

In another embodiment, the way of offset is angular offset, wherein the position of the target line is rotated by angular offset of the specified central line. In yet another embodiment, examples of offsets include position and angle offsets, where the position of the target line is shifted by first shifting the central line according to the positional offset, and then offsetting the displaced central line according to the angle. Move to rotate.

In one embodiment, the lateral control processor 1004 further calculates an equation representing the target line at block 1002 . Then, the lateral control processor 1004 obtains the solution of these equations and the equation (x ² +y ² =d ² ) of the look-ahead distance, and estimates the position of the target point. In another embodiment, the module 1002 directly outputs the above-mentioned target line determination value. Then, module 810 calculates the equation of the target line and estimates the position of the target point. In the above two embodiments, the equation of the target line is determined by the road data (as shown in FIG. 8 ) provided by the road detection device. Due to different driving actions, the target points are correspondingly located on different target lines. The steering actuator then uses the generated steering command to drive the vehicle to perform different driving actions.

Although the present invention is disclosed above with the foregoing embodiments, it is not intended to limit the present invention. Any person skilled in the art may make some changes and modifications without departing from the spirit and scope of the present invention. Therefore, The protection scope of the claims of the present invention shall be defined by the protection scope of the application claims attached to this specification.

Claims (18)

1. a kind of rotating direction control method, be used in a vehicle, the vehicle has and makes the start that multiple tires of the vehicle turn to Device, it is characterised in that the rotating direction control method includes:

Obtain the position for being located remotely from the one of the vehicle target point for seeing front distance；

A progressive position of the vehicle is predicted, the wherein progressive position is disposed far from the sight front distance of the vehicle；

Determine the range difference between the position of the target point and the progressive position；

One normalization errors are calculated by the range difference divided by the sight front distance；And

Determine that a course changing control instructs according to the integration of the normalization errors, wherein the actuator makes according to course changing control instruction The multiple tire turns to, so that the vehicle performs different traveling actions.

2. rotating direction control method as claimed in claim 1, it is characterised in that the sight front distance be according at least with the next item down because Element and determine：Speed, vehicle yaw speed, vehicle lateral acceleration, vehicle steering angle, track camber, the vehicle to track The distance in sideline, the distance of the vehicle to barrier, the normalization errors, the range difference and vehicle traveling action.

3. rotating direction control method as claimed in claim 1, it is characterised in that the sight front distance is the linear function of speed.

4. rotating direction control method as claimed in claim 1, it is characterised in that the position of the target point is held according to the vehicle What capable traveling was acted and positioned, wherein：

When the vehicle is performing track holding action, the target point is in the track that the vehicle is travelled, and offsets the car One deviation post of the Central Line in road, and including zero offset；When the vehicle is performing lane changing action, the target point is position In the track that the vehicle to be converted, a deviation post of the Central Line in the track is offset, and including zero offset；When the vehicle When performing or right-hand rotation acts, the target point is located in the vehicle track to be transferred to, and offsets the center in the track One deviation post of line, and including zero offset；And

The position of the target point is calculated according to the Central Line in the track, the deviation post and the sight front distance.

5. rotating direction control method as claimed in claim 1, it is characterised in that estimating for the progressive position is that basis works as the vehicle When travelling the sight front distance, it is assumed that the vehicle keeps speed instantly and deflection speed instantly, and deflection speed instantly is Determined according at least one of following：The yaw rate of vehicle instantly, and the steering angle of vehicle instantly.

6. rotating direction control method as claimed in claim 1, it is characterised in that estimating for the progressive position is that basis works as the vehicle When travelling the sight front distance, it is assumed that the vehicle keeps speed instantly and steering angle instantly.

7. a kind of rotating direction control method, be used in a vehicle, the vehicle has and makes the start that multiple tires of the vehicle turn to Device, it is characterised in that the rotating direction control method includes:

Traveling action according to performed by the vehicle is taken at a score of the vehicle front；

A progressive position of the vehicle is estimated, the wherein progressive position is disposed far from the one of the vehicle and sees front distance；

Determine the progressive position to a range difference of the score；

One normalization errors are calculated by the range difference divided by the sight front distance；And

Determine that a course changing control instructs according to the integration of the normalization errors, wherein the actuator makes according to course changing control instruction The multiple tire turns to, so that the vehicle performs traveling action.

8. rotating direction control method as claimed in claim 7, it is characterised in that the sight front distance be according at least with the next item down because Element and determine：Speed, vehicle yaw speed, vehicle lateral acceleration, vehicle steering angle, track camber, the vehicle to track The distance in sideline, the distance of the vehicle to barrier, the normalization errors, the range difference and vehicle traveling action.

9. rotating direction control method as claimed in claim 7, it is characterised in that the position of the score is held according to the vehicle The position of the capable traveling operating position fixing score, wherein：

When the vehicle is performing track holding action, the score is in the track that the vehicle is travelled, and offsets the car One deviation post of the Central Line in road, and including zero offset；When the vehicle is performing lane changing action, the score is position In the track that the vehicle to be converted, a deviation post of the Central Line in the track is offset, and including zero offset；When the vehicle When performing or right-hand rotation acts, the score is located in the vehicle track to be transferred to, and offsets the center in the track One deviation post of line, and including zero offset；And

The position of the score is calculated according to the Central Line in the track and the deviation post.

10. a kind of crosswise joint system, the crosswise joint system is installed on the vehicle with multiple tires, to control the vehicle Steering, it is characterised in that the crosswise joint system includes：

One road detection apparatus, for providing the road data of the vehicle front；

One speed sensor, for providing GES；

One steering angle inductor, for providing steering angle signal；

One processor, the processor connect the road detection apparatus to receive track data, and the processor is also connected with the speed sense Device is answered to receive the GES, the processor also connects the steering angle inductor to receive the steering angle signal, at this Manage device and calculate a steering angle order according to following steps：Determine a sight front distance, calculate before the sight of vehicle front away from From a target point position, predict a vehicle progressive position, calculate between the aiming spot and the vehicle progressive position One range difference, and a normalization errors are calculated by the range difference divided by the sight front distance, the normalization errors are integrated；With And

At least one turns to actuator, and the steering actuator turns to the multiple tire according to the steering angle order, so that should Vehicle performs traveling action.

11. crosswise joint system as claimed in claim 10, it is characterised in that the road detection apparatus includes a video sensing Device and an image process unit, the image sensor catch the image of the vehicle front, and the image process unit is according to the image The road data for the image that inductor is caught calculates road shape.

12. crosswise joint system as claimed in claim 10, it is characterised in that the road detection apparatus includes being used to determine one A satellite navigation system, a numerical map and a processing unit for vehicle location, the processing unit build on the vehicle location In the numerical map, and provide the road data in front of the vehicle location.

13. crosswise joint system as claimed in claim 10, it is characterised in that the road detection apparatus scans including a laser Device and a processing unit, the laser scanner send laser pulse wave, and catch the light beam that the object in the vehicle front is reflected, The processing unit determines road shape according to the road data of the reflected light.

14. crosswise joint system as claimed in claim 10, it is characterised in that the processor is according at least one of following factor Determine the sight front distance：Speed, vehicle yaw, vehicle lateral acceleration, vehicle steering angle, track camber, the vehicle are extremely The row that the distance in track sideline, the distance of the vehicle to barrier, the normalization errors, the range difference and vehicle are carrying out Sail action.

15. crosswise joint system as claimed in claim 10, it is characterised in that the processor is carried according to road detection apparatus The road data of confession estimates a lane center line, and the position of the target point is calculated according to the lane center line and the sight front distance Put, wherein the target point is located at a deviation post of the offset lanes Central Line of the sight front distance of the vehicle front, and including zero Skew.

16. crosswise joint system as claimed in claim 10, it is characterised in that the processor pushes away according to the steering angle signal Estimate a vehicle yaw speed, the processor calculates the car further according to the GES, the vehicle yaw speed and the sight front distance Progressive position.

17. crosswise joint system as claimed in claim 10, it is characterised in that further include a deflection speed inductor, be used for A vehicle yaw rate signal is provided, the GES that wherein processor is provided according to the speed sensor, the deflection Vehicle yaw rate signal and the sight front distance that speed sensor is provided predict the vehicle progressive position.

18. crosswise joint system as claimed in claim 10, it is characterised in that the crosswise joint system, which is further connected to, to be installed on The vehicle traveling action decision package of the vehicle, the wherein processor receive what is sent by vehicle traveling action decision package One traveling action command, the traveling action command include at least one of following instruction：Track holding instructs, lane changing instructs, Turn left instruction, instruction of turning right, and barrier is dodged instruction, the processor according to the traveling action command one score of selection, Wherein the target point is located at the score, and the processor calculates the position of the target point according to the score and the sight front distance.