CN113085853B - An assisted driving system for actively dodging large vehicles in the lane - Google Patents

Abstract

The invention provides an assisted driving system for actively dodging large vehicles in a lane, comprising a perception module, a decision module, a planning module, a tracking module and a controller, the perception module includes an environment perception sensor, and the signal of the environment perception sensor is connected to the decision module The input terminal of the decision module determines the four states of the assisted driving system through the signal of the environmental perception sensor. The output terminal of the decision module is connected to the input terminal of the planning module. The planning module is used to plan the vehicle motion trajectory of the two processes of deviation and regression. The module output signal is connected to the input of the tracking module. The assisted driving system for actively dodging large vehicles in the lane according to the present invention can realize automatic target recognition, decision-making, active dodging trajectory planning and dodging action execution functions in the scene of overtaking large vehicles, and can improve the vehicle's performance in this scene. Side-impact safety, and the peace of mind of the driver and occupant.

Description

An assisted driving system for actively dodging large vehicles in the lane

technical field

The invention belongs to the technical field of active safety technology, advanced driving assistance system, perception, decision-making, planning and tracking technology of automatic driving, and in particular relates to an auxiliary driving system for actively dodging large vehicles in a lane.

Background technique

ADAS (Advanced Driver Assistant System) is an advanced driver assistance system that collects road and environmental vehicle parameters in the vehicle's driving environment through on-board sensors, and performs target identification, detection and tracking, so as to predict the risks that the vehicle may encounter and actively change the vehicle. the movement status of the vehicle or proactively alert the driver to improve safety. Sensors commonly used in ADAS systems include: millimeter-wave radar, lidar, ultrasonic radar, cameras, etc. According to the actuators controlled by its assisted driving system, ADAS systems can be divided into ACC (adaptive cruise), FCW (forward collision warning), etc. mainly for longitudinal motion control, LKA (lane keeping) for lateral motion control, LDW (Lane Departure Warning), ALC (Automatic Lane Change), etc., as well as TJA (Traffic Jam Assist), HWP (High Speed Autonomous Driving) and other functions that combine the two;

Among them, among the LKA, LDW, ALC and other functions for planning and controlling lateral motion, the purpose of LKA and LDW is to control the vehicle to keep driving at the center of the lane line, which belongs to the lateral motion control in the own lane, while ALC It is based on the perception information to judge the lane change intention and control the cross-lane lateral motion control of the vehicle to change lanes;

For the scenario where the ego vehicle overtakes a large vehicle, due to the large vehicle width, large lateral position fluctuation and obvious negative pressure area at the rear of the car, if the ego vehicle LKA function is activated, the ego vehicle will continue to keep the center line of the lane. , it will lead to a small distance between the self-vehicle and the large vehicle during the overtaking process, there is a risk of side collision, and it is easy to bring a bad driving experience to the drivers and passengers. The introduction of the DWEL system in the present invention can fundamentally solve this problem. When the self-vehicle sensor detects the target scene of overtaking a large vehicle, the DWEL system is turned on to control the self-vehicle to deviate in the opposite direction (inside the self-lane) in advance when approaching the target, and control the self-vehicle to return to the centerline of the lane after the overtaking is completed. . The introduction of the DWEL system not only ensures the stability and comfort of the vehicle, but also increases the distance between vehicles based on environmental parameters, thereby improving the side collision safety during overtaking and improving the driver's "safety". At the same time, the trajectories of the deviation process and the regression process are planned based on real-time environmental parameters and driving behavior maps, which conform to the driving habits of human drivers and have the characteristics of safe anthropomorphism.

SUMMARY OF THE INVENTION

In view of this, the present invention aims to propose an assisted driving system (DWEL, Dodge Within Ego Lane) that actively avoids large vehicles in the lane, so as to solve the risk of side collision when the LKA function of the self-vehicle is activated, and the driver and passengers are not allowed to drive. problem of experience.

In order to achieve the above object, the technical scheme of the present invention is achieved in this way:

An assisted driving system for actively dodging large vehicles in a lane, comprising a perception module, a decision module, a planning module, a tracking module and a controller, the perception module includes an environment perception sensor, the environment perception sensor performs perception operations on the environment, and detects the target level The information is acquired, the target-level information includes the target vehicle identification and environmental parameters, and the target-level information is processed, and then the processed target-level information is sent to the decision-making module, and the decision-making module receives the processed target-level information. The machine condition transfer list is used to convert the four states of the assisted driving system, and determine the current state of the assisted driving system. The four states of the assisted driving system include the activation state. The information is processed to obtain decision-making command parameters, and then the decision-making command parameters are sent to the planning module. The planning module receives the decision-making command parameters, and the planning module uses the polynomial trajectory planning method to plan the self-vehicle to avoid the deviation of large vehicles according to the current environmental parameters and driving behavior map. The planned trajectory parameters are obtained from the trajectory and the regression trajectory, and the planned trajectory parameters are sent to the tracking module. The tracking module receives the planned trajectory parameters, and controls the actuator of the vehicle to ensure that the vehicle moves along the target trajectory. The environment perception sensor, The decision-making module, the planning module and the tracking module are all connected with the controller by signals.

Further, the environment perception sensor includes a camera, a steering wheel torque sensor, a steering wheel angle sensor, an accelerator pedal sensor, a brake pedal sensor and a gyroscope. The pedal sensor, brake pedal sensor and gyroscope all have signals connected to the controller.

Further, the four states of the assisted driving system also include an off state, a standby state and a fault state, the off state is turned on and enters the standby state, the standby state is turned off and enters the off state, the standby state is activated and enters the active state, and the activated state exits and enters the standby state. , when the system is in the standby state and active state, if a fault is detected, it will enter the fault state, and the fault state will enter the off state when the fault is eliminated.

Further, the processing operation of the target-level information includes the following steps:

A1. The perception module determines whether the target vehicle type is a large vehicle, if so, go to the next step, otherwise switch to the next target vehicle and re-enter step A1;

A2. The perception module determines whether the longitudinal distance of the target vehicle relative to the vehicle is within the threshold ΔDx (150m), if so, proceed to the next step, otherwise switch to the next target vehicle and re-enter step A1;

A3. The perception module judges whether the lateral distance of the target vehicle relative to the vehicle is within the range of ΔDy ₁ (2-5m). If so, the Left_Count increases by 1, switches to the next target vehicle, and re-enters step A1, otherwise, proceed to the next step;

A4. The perception module judges whether the lateral distance of the target vehicle relative to the vehicle is within the range ΔDy ₂ (-5～-2m), if so, the Right_Count increases by 1, switches to the next target vehicle, and re-enters step A1, otherwise switches to the next target vehicle , and re-enter step A1;

A5. When all target vehicles have gone through the judgment process of A1~A4, the perception module judges whether Left_Count>0 and Right_Count=0, if Left_Count>0 and Right_Count=0, the target vehicle is on the left side, and the output relative longitudinal distance is the smallest All the information of the large vehicle, meet the requirements of the target scene at this time, otherwise go to the next step;

A6. The perception module judges whether Left_Count = 0 and Right_Count > 0. If Left_Count = 0 and Right_Count > 0, the target vehicle is located on the right side, and outputs all the information of the large vehicle with the smallest relative longitudinal distance. At this time, it meets the requirements of the target scene. Otherwise, go to the next step;

A7. The perception module judges whether Left_Count = 0 and Right_Count = 0, if Left_Count = 0 and Right_Count = 0, there is no large target vehicle, and the DWEL system is in a standby state, otherwise, go to the next step;

A8. The perception module judges whether Left_Count>0 and Right_Count>0, if Left_Count>0 and Right_Count>0, there are large vehicles on both sides of the road, and the DWEL system is in a standby state.

Further, the judging process in step A5 includes four kinds of possibility judgments, and the four kinds of possibility judgments respectively include only going through step A1, going through step A1, step A2 in turn, going through step A1, step A2, step A3 in turn, Go through Step A1, Step A2, Step A3, and Step A4 in sequence.

Further, the polynomial trajectory planning method of the deviation trajectory and the regression trajectory includes the following parameters: TTO (Time To Overtake), τ ₁ , τ ₂ , Δy, TTO calculation formula, τ ₁ and τ ₂ parameter setting method, Δy calculation formula as follows:

TTO=Δx/(u _SV -u _target );

τ ₁ =0.0394·(u _SV -u _target )+3.3159(s);

τ ₂ =-0.0298·(u _SV -u _target )+4.6306(s);

Among them, TTO refers to the time when the ego car and the target car drive at a constant speed at the current speed, and the time until the ego car overtakes the target car, Δx is the relative longitudinal distance between the front of the ego car and the rear of the target car, u _sv is the current speed of the ego car, u _target is the current speed of the target vehicle. τ ₁ and τ ₂ are the durations of the departure process and the regression process, respectively, Δy is the maximum offset of the departure process, and x ₁ and x ₂ are the lateral distances between the ego vehicle and the target large vehicle (the gap between the two vehicle bodies), respectively. ) and the distance between the ego vehicle and the side lane line of the target large vehicle.

Further, the control operation of the tracking module includes the following steps:

B1. The tracking module uses PID control to track the target trajectory,

B2. The tracking module introduces particle swarm optimization (PSO) and preview model to achieve adaptive tuning PID control;

B3. The tracking module obtains the optimal position through continuous iterative operations on all particle states.

Compared with the prior art, the assisted driving system for actively dodging large vehicles in a lane according to the present invention has the following advantages:

(1) The assisted driving system for actively dodging large vehicles in the lane according to the present invention can realize automatic target recognition, decision-making, active dodging trajectory planning and dodging action execution functions in the scene of overtaking large vehicles, and can improve the performance of vehicles in this area. The side collision safety in the scene, as well as the peace of mind of the driver and passengers.

(2) The planning strategy formulated by the assisted driving system for actively dodging large vehicles in the lane according to the present invention in the dodging trajectory planning process is based on the anthropomorphic development of natural driving behavior data, which can ensure that this function conforms to human (Chinese). ) driving habits.

(3) The dodging action execution function of the assisted driving system for actively dodging large vehicles in the lane according to the present invention can be developed on the basis of the underlying actuator of the LKA function, and the required reconstruction cost is low.

Description of drawings

The accompanying drawings constituting a part of the present invention are used to provide further understanding of the present invention, and the exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the attached image:

1 is a schematic diagram of a camera detection target of an assisted driving system for actively dodging large vehicles in a lane according to an embodiment of the present invention;

FIG. 2 is a flow chart of target type determination in the perception link of an assisted driving system for actively dodging large vehicles in a lane according to an embodiment of the present invention;

3 is a state machine transition diagram of an assisted driving system for actively dodging large vehicles in a lane according to an embodiment of the present invention;

4 is a schematic diagram of the trajectory planning of the deviation and return process of an assisted driving system for actively dodging large vehicles in a lane according to an embodiment of the present invention;

5 is a flow chart of trajectory tracking of an assisted driving system for actively dodging large vehicles in a lane according to an embodiment of the present invention;

FIG. 6 is a flow chart of optimization of a particle swarm optimization algorithm of an assisted driving system for actively dodging large vehicles in a lane according to an embodiment of the present invention;

FIG. 7 is a result diagram of an adaptive tuning PID trajectory tracking test based on the PreScan-Simulink simulation platform of an assisted driving system for actively dodging large vehicles in a lane according to an embodiment of the present invention;

8 is a schematic diagram of the error of the adaptive tuning PID trajectory tracking test result based on the PreScan-Simulink simulation platform of an assisted driving system for actively dodging large vehicles in a lane according to an embodiment of the present invention.

Detailed ways

It should be noted that the embodiments of the present invention and the features of the embodiments may be combined with each other under the condition of no conflict.

In the description of the present invention, it should be understood that the terms "center", "portrait", "horizontal", "top", "bottom", "front", "rear", "left", "right", " The orientation or positional relationship indicated by vertical, horizontal, top, bottom, inner, outer, etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and The description is simplified rather than indicating or implying that the device or element referred to must have a particular orientation, be constructed and operate in a particular orientation, and therefore should not be construed as limiting the invention. In addition, the terms "first", "second", etc. are used for descriptive purposes only, and should not be construed as indicating or implying relative importance or implying the number of indicated technical features. Thus, a feature defined as "first", "second", etc., may expressly or implicitly include one or more of that feature. In the description of the present invention, unless otherwise specified, "plurality" means two or more.

In the description of the present invention, it should be noted that the terms "installed", "connected" and "connected" should be understood in a broad sense, unless otherwise expressly specified and limited, for example, it may be a fixed connection or a detachable connection Connection, or integral connection; can be mechanical connection, can also be electrical connection; can be directly connected, can also be indirectly connected through an intermediate medium, can be internal communication between two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood through specific situations.

The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

As shown in Figure 1, an assisted driving system (ie DWEL) for actively dodging large vehicles in a lane includes a perception module, a decision module, a planning module, a tracking module and a controller. The sensor signal is connected to the input terminal of the decision module, the decision module determines the four states of the assisted driving system through the environmental perception sensor signal, the output terminal signal of the decision module is connected to the input terminal of the planning module, and the planning module is used for planning deviation and regression. The vehicle motion trajectory of the process, the signal of the output end of the planning module is connected to the input end of the tracking module, the decision-making module, the planning module, and the tracking module are all signal-connected to the controller, and the controller is an ECU. The vehicle's risk scene, actively perceive the surrounding environment, and plan the self-vehicle to avoid the deviation trajectory and return trajectory of large vehicles according to the current environmental parameters and driving behavior map, so as to ensure appropriate vehicle clearance during the overtaking process and improve driving safety.

The environmental perception sensor includes a camera, a steering wheel torque sensor, a steering wheel angle sensor, an accelerator pedal sensor, a brake pedal sensor and a gyroscope, the camera, the steering wheel torque sensor, the steering wheel angle sensor, the accelerator pedal sensor, Both the brake pedal sensor and the gyroscope are signally connected to the controller, and the camera perception information includes but is not limited to: object ID, type, relative distance, relative speed, left and right lane line information, and the like. The steering wheel torque sensor is used to detect the torque applied by the driver to the steering wheel. The steering wheel angle sensor is used to detect the steering wheel angle. The accelerator pedal sensor is used to detect the accelerator pedal opening. The brake pedal sensor is used to detect the brake pedal angle. The gyroscope is used to detect the vehicle speed and longitudinal and lateral accelerations. If the target within a certain longitudinal range satisfies the existence of large vehicles only in the one-sided adjacent lane (left adjacent lane or right adjacent lane) of the vehicle, the current environment meets the target scene conditions of the DWEL system, otherwise, the DWEL system is not turned on .

As shown in Figure 3, the four states of the assisted driving system include OFF state (OFF), standby state (STANDBY), active state (ACTIVE) and fault state (FAILURE), and the DWEL system decision module determines these four states conversion between. The off state is turned on and enters the standby state, the standby state is turned off and enters the off state, and the standby state is activated and enters the active state. The activation state includes the deviation process and the return process. If the deviation process and the return process are completed, the DWEL system exits and enters the standby state. When the system is in the standby state and active state, if a fault is detected, it will enter the fault state, and the fault state will enter the shutdown state when the fault is eliminated.

The assisted driving system includes the following operation steps:

S1. The environmental perception sensor performs a perception operation on the environment, identifies the target vehicle and obtains environmental parameters, and processes the target-level information, and then transmits the processed target-level information to the decision-making module; identification of the target vehicle and environmental parameters The acquisition includes relative motion parameters of the target vehicle, lane line information, etc.;

S2. The decision-making module receives the processed target-level information, converts the four states of the DWEL system according to the state machine conditional transition list, and determines the current state of the DWEL system. In the active state, the decision-making module processes the processed target-level information. Perform processing operations to obtain decision-making command parameters, and then send the decision-making command parameters (referring to the decision-making command parameters output by the decision-making module) to the planning module; the current state of the DWEL system includes four types: shutdown, standby, activation, and failure;

S3. The planning module receives the decision command parameters, and the planning module plans the deviation trajectory and the return trajectory of the self-vehicle to avoid the large vehicle by using the polynomial trajectory planning method according to the current environmental parameters and the driving behavior map, and transmits the planned trajectory parameters to the tracking module;

S4. The tracking module receives the planned trajectory parameters (planned trajectory parameters output by the planning module), and controls the actuator of the vehicle to ensure that the vehicle moves along the target trajectory. The actuator of the vehicle is the steering-by-wire system, and the DWEL system can pass The camera performs target type recognition and lateral distance detection on large vehicles encountered during driving. Its working process can be divided into two stages: 1 Deviation stage: when the self-vehicle approaches the target Calculate the target lateral offset Δy, deviation time τ ₁ and return time τ ₂ based on information such as lane line distance, ego vehicle speed, and target longitudinal relative speed, and send the bottom actuator to steer to achieve the deviation to maintain a lateral safety distance; 2 Regression phase: Re-return to the lane centerline position after overtaking the target. Based on ACC, LKS, TJA, HWP and other systems, this function can further improve the lateral safety of the vehicle and the reassurance of drivers and passengers.

In addition, the core of the decision-making module is the state machine. As shown in Figure 3, it includes four states: OFF, STANDBY, ACTIVE, and FAILURE. The initial state of the decision-making module should be in the OFF state, and when the driver turns on the DWEL system, it will enter the standby state. , when the perception module processes the target-level information and determines that it meets the requirements of the target scene, it enters the active state, and if the decision module is in the standby state, the driver turns off the DWEL system and enters the off state. If and only when the decision-making module is in the standby state, the planning module plans the deviation trajectory and the return trajectory of the self-vehicle dodging the large vehicle. After the self-vehicle regression phase is completed, the decision-making module exits from the active state and enters the standby state. When the decision-making module is in the active state or standby state and the system detects that the sensing module, decision-making module, planning module or tracking module is faulty, the decision-making module reports an error and enters the fault state, until the system detects that the fault is eliminated, and the decision-making module enters the closed state.

As shown in Figure 2, the processing operation of the target-level information in step S1 includes the following steps:

A1. The perception module judges whether the target vehicle type is a large vehicle (truck or bus), if so, go to the next step; otherwise, switch to the next target vehicle and re-enter step A1;

A2. The perception module determines whether the longitudinal distance of the target vehicle relative to the vehicle is within the threshold ΔDx (150m), if so, proceed to the next step, otherwise switch to the next target vehicle and re-enter step A1;

A3. The perception module judges whether the lateral distance of the target vehicle relative to the vehicle is within the range of ΔDy ₁ (2-5m). If so, the Left_Count increases by 1, switches to the next target vehicle, and re-enters step A1, otherwise, proceed to the next step;

A4. The perception module judges whether the lateral distance of the target vehicle relative to the vehicle is within the range ΔDy ₂ (-5～-2m), if so, the Right_Count increases by 1, switches to the next target vehicle, and re-enters step A1, otherwise switches to the next target vehicle , and re-enter step A1;

A5. When all target vehicles have gone through the judgment process of A1~A4, the perception module judges whether Left_Count>0 and Right_Count=0, if Left_Count>0 and Right_Count=0, the target vehicle is located on the left side, and the output relative longitudinal distance is the smallest All the information of the large vehicle, meet the requirements of the target scene at this time, otherwise go to the next step;

A6. The perception module judges whether Left_Count = 0 and Right_Count > 0. If Left_Count = 0 and Right_Count > 0, the target vehicle is located on the right side, and outputs all the information of the large vehicle with the smallest relative longitudinal distance. At this time, it meets the requirements of the target scene. Otherwise, go to the next step;

A7. The perception module judges whether Left_Count = 0 and Right_Count = 0, if Left_Count = 0 and Right_Count = 0, there is no large target vehicle, and the DWEL system is in a standby state, otherwise, go to the next step;

A8. The perception module judges whether Left_Count>0 and Right_Count>0, if Left_Count>0 and Right_Count>0, there are large vehicles on both sides of the road, and the DWEL system is in a standby state.

It should be noted that the main design object of the assisted driving system for actively dodging large vehicles in this lane is the lateral decision-making planning and control of the vehicle. When the self-vehicle speed is low, it cannot overtake the large vehicle on the left. At this time, the DWEL system is in an active state. According to the formula and definition of TTO and τ ₁ of the planning module, it can be known that when the system enters the active state, it starts to time, and after the time ( After TTO-τ ₁ ), it begins to enter the deviation process. If the speed of the ego vehicle is lower than the speed of the large vehicle in the left lane, TTO < 0, meaningless, the system will continue to remain active but will not perform the deviation action until the target vehicle drives out of the perception threshold range, the system returns to the standby state.

The judgment process in step A5 includes four kinds of possibility judgments, and the four kinds of possibility judgments respectively include only going through step A1, going through step A1, step A2 in turn, going through step A1, step A2, step A3 in turn, going through steps in turn A1, step A2, step A3, step A4.

In the actual test, after obtaining the target-level information of all target vehicles, assuming that there are 5 target vehicles, the first target vehicle first goes through step A1 to determine whether it is a large vehicle. If it is a large vehicle, the first target vehicle enters the step A2, if it is not a large vehicle, the first target vehicle stops at step A1 and waits for the other four target vehicles to go through the judgment process;

At the same time, switch to the second target vehicle and enter step A1 to determine whether it is a large vehicle. If it is a large vehicle, the second target vehicle enters step A2 (if it is not a large vehicle, the second target vehicle stops at Step A1, wait for the remaining three target vehicles to go through the judgment process), if the second target vehicle is in step A2, and its longitudinal distance relative to the vehicle is within the threshold ΔDx (20m), then the second target vehicle enters step A3, If the longitudinal distance of the second target vehicle relative to the own vehicle is not within the threshold ΔDx (20m), the second target vehicle stops at step A2 and waits for the remaining three target vehicles to complete the judgment process;

The judging process of the third target car, the fourth target car, and the fifth target car is analogized in turn. When all the target cars have gone through the judgment process of steps A1 to A4, the next step can be performed in step A5.

As shown in FIG. 4 , the polynomial trajectory planning method of the deviation trajectory and regression trajectory in step S3 includes the following parameters: TTO (Time To Overtake), τ ₁ , τ ₂ , Δy, TTO calculation formula, τ ₁ and τ ₂ The parameter setting method and Δy calculation formula are as follows:

TTO=Δx/(u _SV -u _target );

τ ₁ =0.0394·(u _SV -u _target )+3.3159(s);

τ ₂ =-0.0298·(u _SV -u _target )+4.6306(s);

Among them, TTO refers to the time when the ego car and the target car are driving at a constant speed at the current speed, and the time until the ego car overtakes the target car, Δx is the relative longitudinal distance between the ego car’s front and the target car’s rear, that is, the longitudinal distance between the two cars, u _sv is the current speed of the ego vehicle, and u _target is the current speed of the target vehicle. τ ₁ and τ ₂ are the durations of the departure process and the regression process, respectively, Δy is the maximum offset of the departure process, and x ₁ and x ₂ are the lateral distances between the ego vehicle and the target large vehicle (the gap between the two vehicle bodies), respectively. the lateral length) and the distance between the vehicle and the side lane line of the target large vehicle, in addition, the three parameters τ ₁ , τ ₂ , Δy are not constant values, but change in real time according to the surrounding environment of the vehicle, such as : τ ₁ will increase with the increase of the longitudinal speed difference between the ego vehicle and the target large vehicle, and τ ₂ will decrease with the increase of the longitudinal speed difference between the ego vehicle and the target large vehicle. The parameters of τ ₁ and τ ₂ are set It can be set based on driving behavior map, empirical formula, etc.: After the system enters the active state, TTO is calculated in real time. When TTO is less than τ ₁ and TTO > 0s, the deviation process is started, and the regression process is entered after τ ₁ time, and the regression process continues. After τ ₂ , the DWEL system exits the active state.

The trajectory planning part of the deviation process and the regression process is carried out based on the polynomial trajectory planning method. The maximum offset Δy of the deviation process in the present invention is jointly determined according to environmental parameters and expert experience. It decreases with the increase of the distance x ₁ (the lateral length of the gap between the two vehicle bodies), and decreases with the increase of the distance x ₂ between the vehicle and the side lane line of the target large vehicle, as shown in the figure 4 shown.

On the basis of the above, the initial and final state boundary conditions of the dodging trajectory are introduced as follows, and the polynomial parameters can be solved according to the polynomial trajectory planning model:

y ₁ (t)=a ₀ +a ₁ t+a ₂ t ² +a ₃ t ³ ;

y _l represents the lateral trajectory of the vehicle, a ₀ to a ₃ are the coefficients of the lateral trajectory planning model designed based on the cubic polynomial trajectory planning method, and have no physical meaning. The calculation method has been listed in the formula, and Δy is the deviation process The maximum offset of , t represents the duration of the deviation process.

As shown in Figure 5, the tracking module control operation in step S4 includes the following steps:

B1. The tracking module uses PID control to track the target trajectory,

B2. The tracking module introduces particle swarm optimization (PSO) and preview model to achieve adaptive tuning PID control;

B3. The tracking module obtains the optimal position (optimal solution of the optimization problem) through continuous iterative operation of all particle states; the input of the trajectory tracking module of the DWEL system is the target trajectory output by the planning module, and the output is the steering wheel angle. The trajectory tracking control process is shown in Figure 5. The PID control is used to track the target trajectory. On the basis of the traditional PID control, the particle swarm optimization algorithm (PSO) and the preview model are introduced to realize the adaptive tuning of the PID. control to reduce the error and delay of PID control. Among them, particle swarm optimization is a kind of swarm intelligence algorithm used to solve the objective optimization problem, which has flexibility, robustness and self-organization. The core idea of particle swarm optimization is to build a population composed of several particle individuals in the solution space of the optimization problem. The initial state of the particles is random and can move freely in the space, and all particles in the population are iteratively optimized to achieve the optimal problem. solution. The space state of each particle can be expressed as x _i =[x _i1 ,x _i2 ...... x _iD ], the dimension D of each particle is equal to the dimension of the solution space (D=3 in this patent, the solution dimension three control parameters for pid), and each particle has four characteristics: particle space position p, particle motion velocity vector v, fitness value fitness and particle’s individual extreme value g. Among them, the particle space position represents the possible optimal solution of the optimization problem, the particle motion velocity vector represents the optimization direction and gradient of the optimal solution, the fitness value represents the mapping value of each particle to the fitness function, and the individual extreme value represents each The position where the particle is closest to the optimal solution of the model during optimization. In addition, there is another optimization parameter z at the population level: the population extremum, which represents the individual extremum that is closest to the optimal solution in each iteration of the population. During each iteration of the optimization process, the update formulas for particle velocity and position are as follows:

where w is the inertia weight, d∈[1,D], rand ₁ and rand ₂ are two random factors between 0 and 1, respectively, and c ₁ and c ₂ are acceleration factors. The variation interval of particle position and velocity can be set artificially as [p _min p _max ] and [v _min v _max ]. The power of particle motion comes from three aspects: (1) inertial force, which keeps the particle moving in the initial direction to achieve global search; (2) individual extreme attraction, which guides the particle to move towards its own historical optimal position and maintain it at this position to achieve self-realization cognition; ③ population extreme attractiveness, guiding particles away from the initial motion direction and moving towards the historical optimal position of other particles to realize population cognition. The particle swarm optimization process is shown in Figure 6.

In order to realize the adaptive tuning PID control, the optimal p, i, d control parameters can be optimized based on the particle swarm algorithm (vehicle speed, lateral distance of dodging, deviation process time) under certain initial conditions, where the particle state can be expressed as is X _i = [K _p K _i K _d ], and the fitness value corresponding to each particle is obtained from the simulation test results. The fitness value fitness consists of three parts: the actual lateral displacement and the The lag distance e ₁ of the target lateral displacement, the lag distance e ₂ between the actual lateral displacement and the target lateral displacement at the end of the deviation process, and the maximum overshoot e ₃ of the lateral displacement between the actual trajectory and the target trajectory after the regression process ends. After obtaining the optimal PID parameters under different initial working conditions, a PID truth table is formulated, and different PID parameters are selected under different initial working conditions (vehicle speed, lateral distance for dodging, and deviation process time), so as to achieve different working conditions. The adaptively tunable trajectory tracking algorithm under the conditions improves the robustness of the trajectory tracking module.

fitness=e ₁ +e ₂ +e ₃ ;

e ₁ =|y _target (t)-y _real (t)||t=t ₀ +1s;

e ₂ =|y _target (t)-y _real (t)| |t=t ₀ +τ ₁ ;

e ₃ =|y _target (t)-y _real (t)| |t=t ₀ +τ ₁ +τ ₂ ;

Among them, the vertical line in the formula represents the boundary condition, that is, e ₁ is the error at time t ₀ +1s, e ₂ is the error at time t ₀ +τ ₁ , and e ₃ is the error at time t ₀ +τ ₁ +τ ₂ ; y _target (t) represents the lateral displacement value of the target trajectory output by the planning module, and y _real (t) represents the lateral displacement value of the actual trajectory of the vehicle in the simulation test.

The adaptive tuning PID trajectory tracking system proposed in the present invention is simulated and verified in the PreScan-Matlab-Simulink co-simulation platform. In the scene targeted by the invention, it has high tracking accuracy and can meet the requirements of real vehicle control.

In addition, it should be noted that in FIG. 1, the camera sensor is the camera that appears in the text, the field of view of the camera sensor and the detection range of the camera; OBJ ₁ to OBJ ₄ are the schematic diagrams of the objects detected by the camera, and (x, y) represents the target Object relative longitudinal distance and relative lateral distance;

In Fig. 2, OBJ _i (x _i , y _i ) represents the i-th target vehicle and its relative longitudinal distance and relative lateral distance information;

In Figure 4, the left adjacent lane represents the first lane on the left side of the vehicle, and the right adjacent lane represents the first adjacent lane on the right side of the vehicle;

In Fig. 5, the trajectory tracking error is the difference between the lateral displacement of the target trajectory and the actual trajectory, the control variable is the steering wheel angle, and the controlled object is the dynamic system of the vehicle;

In Figure 6, Y represents that the accuracy judgment is satisfied, and the optimization result is output. N means that the accuracy judgment is not satisfied, then continue to iteratively update the speed and position of the particle.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the scope of the present invention. within the scope of protection.

Claims (6)

1. The utility model provides an active driver assistance system who dodges large vehicle in lane which characterized in that: the system comprises a sensing module, a decision-making module, a planning module, a tracking module and a controller, wherein the sensing module comprises an environment sensing sensor which is used for sensing the environment and acquiring target-level information, the target-level information comprises target vehicle identification and environmental parameters and is processed and then is transmitted to the decision-making module, the decision-making module receives the processed target-level information, the four states of an auxiliary driving system are converted according to a state machine condition transfer list and determines the current state of the auxiliary driving system, the four states of the auxiliary driving system comprise an activated state, the decision-making module processes and operates the processed target-level information to obtain decision-making instruction parameters in the activated state, the decision-making instruction parameters are transmitted to the planning module, and the planning module receives the decision-making instruction parameters, the planning module plans a deviation track and a regression track of the self-vehicle for avoiding the large vehicle through a polynomial track planning method according to the current environment parameters and the driving behavior map to obtain planning track parameters, and transmits the planning track parameters to the tracking module, the tracking module receives the planning track parameters and controls and operates an executing mechanism of the vehicle to ensure that the self-vehicle moves along a target track, and the environment perception sensor, the decision module, the planning module and the tracking module are all in signal connection with the controller;

the target level information processing operation comprises the following steps:

a1, judging whether the type of the target vehicle is a large vehicle by the sensing module, if so, carrying out the next step, otherwise, switching the next target vehicle, and re-entering the step A1;

a2, judging whether the longitudinal distance between the target vehicle and the self vehicle is at the threshold value by the sensing module

In the interior of said container body,

if the value of the target vehicle is 150m, carrying out the next step, otherwise, switching to the next target vehicle, and re-entering the step A1;

a3, perception module judgmentWhether the transverse distance of the target-breaking vehicle relative to the self vehicle is in the range or not

In the interior of said container body,

if the range value is 2 m-5 m, increasing 1 for Left _ Count, switching to the next target vehicle, and re-entering the step A1, otherwise, performing the next step;

a4, judging whether the transverse distance of the target vehicle relative to the vehicle is in the range by the sensing module

In the interior of said container body,

if the range value is-5 m to-2 m, the Right _ Count is increased by 1, the next target vehicle is switched, and the step A1 is entered again, otherwise, the next target vehicle is switched, and the step A1 is entered again;

a5, when all target vehicles are subjected to the judging process of A1-A4, judging whether Left _ Count is greater than 0 and Right _ Count is =0 by a sensing module, if Left _ Count is greater than 0 and Right _ Count is =0, the target vehicles are located on the Left side, all information of the large-sized vehicle with the minimum relative longitudinal distance is output, the requirement of a target scene is met at the moment, and if not, the next step is carried out;

a6, judging whether Left _ Count =0 and Right _ Count >0 by the sensing module, if Left _ Count =0 and Right _ Count >0, locating the target vehicle on the Right side, outputting all information of the large vehicle with the minimum relative longitudinal distance, and at the moment, meeting the requirements of the target scene, otherwise, carrying out the next step;

a7, judging whether Left _ Count =0 and Right _ Count =0 by a perception module, if Left _ Count =0 and Right _ Count =0, a large target vehicle does not appear, and an auxiliary driving system for actively dodging the large vehicle in a lane is in a standby state, otherwise, the next step is carried out;

a8, the sensing module determines whether Left _ Count is greater than 0 and Right _ Count is greater than 0, if Left _ Count is greater than 0 and Right _ Count is greater than 0, large vehicles are present on both sides of the road, and the auxiliary driving system actively avoiding the large vehicles in the lane is in a standby state.

2. An in-lane active large vehicle avoidance pilot system according to claim 1, wherein: the environment perception sensor comprises a camera, a steering wheel torque sensor, a steering wheel corner sensor, an accelerator pedal sensor, a brake pedal sensor and a gyroscope, and the camera, the steering wheel torque sensor, the steering wheel corner sensor, the accelerator pedal sensor, the brake pedal sensor and the gyroscope are all in signal connection with the controller.

3. An in-lane active large vehicle avoidance pilot system according to claim 1, wherein: the four states of the auxiliary driving system further comprise a closing state, a standby state and a fault state, wherein the closing state is opened and enters the standby state, the standby state is closed and enters the closing state, the standby state is activated and enters the activation state, the activation state exits and enters the standby state, and when the system is in the standby state and the activation state, if faults are detected, the system enters the fault state, the fault state is removed, and the system enters the closing state.

4. An in-lane active large vehicle avoidance pilot system according to claim 1, wherein: the judgment process in step a5 includes four possibility judgments, each of which includes going through only step a1, going through step a1, step a2 in order, going through step a1, step a2, step A3 in order, going through step a1, step a2, step A3, step a4 in order.

5. An in-lane active large vehicle avoidance pilot system according to claim 1, wherein: the polynomial trajectory planning method comprises the following parameters: TTO,

、

、

TTO formula,

And

the parameter setting method,

The calculation formula is as follows:

；

τ ₁ ＝ 0.0394 · (u_SV-u_target)+3.3159(s) ；

τ ₂ ＝ -0.0298 · (u_SV-u_target)+4.6306(s) ；

；

wherein TTO refers to the time when the own vehicle and the target vehicle both run at a constant speed according to the current speed and exceed the target vehicle from the own vehicle,

is the relative longitudinal distance between the head of the bicycle and the tail of the target vehicle,

in order to obtain the current speed of the vehicle,

the current speed of the target vehicle is taken;

and

respectively the duration of the deviation process and the regression process,

for the maximum amount of deviation of the deviation process,

and

the distance between the self vehicle and the target large-sized vehicle is the lateral distance between the self vehicle and the target large-sized vehicle.

6. An in-lane active large vehicle avoidance pilot system according to claim 1, wherein: the tracking module control operation comprises the steps of:

b1, the tracking module tracks the target track by PID control,

b2, introducing a particle swarm optimization algorithm and a preview model into the tracking module to realize self-adaptive setting PID control;

b3, the tracking module continuously iterates the states of all the particles to obtain the optimal position.

Images