CN109559528B - Self-perception interactive traffic signal control device based on 3D laser radar - Google Patents

Abstract

The invention belongs to the technical field of traffic signal control, and particularly relates to a self-perception interactive traffic signal control device based on a 3D laser radar; the device mainly comprises a 3D laser radar detection unit, a traffic signal interaction coordination unit and a traffic signal drive control unit; the device realizes the autonomous perception of pedestrian flow and vehicle flow in the inlet direction of the intersection and the automatic extraction of traffic flow parameters with real time, high precision and multiple resolutions through the 3D laser radar sensor, and performs refined traffic signal optimization, coordination and control to improve the traffic efficiency of the intersection and improve the utilization rate of road traffic resources.

Description

A self-aware interactive traffic signal control device based on 3D lidar

technical field

The invention belongs to the technical field of traffic signal control, and in particular relates to a self-perception interactive traffic signal control device based on 3D laser radar.

Background technique

With the continuous development of my country's social economy and urbanization process, the characteristics of cities with more people and less land, more cars and less roads, concentrated facilities, tight land use, and frequent activities determine that the land resources available for transportation in cities are extremely limited. The continuous and rapid development of the city will inevitably continue to expand the city's traffic demand, and the contradiction between urban traffic supply and traffic demand has become increasingly prominent. From the perspective of expanding traffic supply: First, rely on new construction and expansion of urban roads. However, under the limited land resources of the city, it is not realistic to build and expand roads, and the speed of road construction can never keep up with the growth rate of traffic demand. The first is to solve the problem of traffic congestion; the second is to improve the efficiency of urban roads. Using intelligent transportation technology to improve the utilization rate of urban road traffic infrastructure and improve road traffic capacity is the main technical means to solve urban traffic problems at home and abroad. Among them, urban road intersections are the bottleneck of road traffic efficiency. To effectively improve road traffic efficiency, only through real-time, microscopic and accurate traffic flow data and fine-grained control of traffic signals can the road traffic efficiency be effectively improved.

In the current traffic signal control system, traffic information is collected mainly through sensing devices such as coil detectors, radar detectors, video detectors, and GPS floating vehicles. However, although the coil detector can obtain the flow, speed, and occupancy information of a certain section of the road, it can only reflect the running state of the traffic flow near the section, and cannot reflect the traffic state within a certain area; similarly, for the same roadbed Although the traffic information obtained by the Doppler radar detector and video detector of the type detection equipment is slightly different, they can only reflect the traffic flow state information of a certain section of the road. For refined traffic signal control, the current The traffic flow information provided by roadbed detectors is not fine enough, and the resolution and dimension of the information cannot meet the requirements of refined traffic optimization control. It needs to be supplemented and improved by installing other detection equipment; although GPS floating vehicle data can reflect Traffic flow information within a certain spatial range, but limited by sample size, sampling period, network transmission and other conditions, the traffic flow information provided by GPS floating vehicles has obvious lag and accuracy fluctuations, and is not suitable for fine-grained traffic signals in control.

As an active vision sensor, 3D lidar has the advantages of insensitivity to external light changes, strong adaptability to complex environments, strong anti-interference ability, high sensitivity, high resolution, high precision, wide coverage, and large amount of information. Real-time, microscopic, high-precision, high-resolution traffic flow information provides new technical solutions for refined traffic information perception, traffic flow dynamic identification and tracking of urban traffic signal control.

Contents of the invention

The present invention proposes a self-perception interactive traffic signal control device based on 3D laser radar. The device is mainly composed of a 3D laser radar detection unit, a traffic signal interaction coordination unit, and a traffic signal drive control unit, as shown in FIG. 1 . The device uses a 3D lidar sensor to realize the independent perception of the flow of people and vehicles in the direction of the intersection entrance and the real-time, high-precision, and multi-resolution automatic extraction of traffic flow parameters, and conduct refined traffic signal optimization, coordination, and control to improve the intersection Improve traffic efficiency and improve the utilization rate of road traffic resources.

The technical solution of the present invention is to install a self-sensing interactive traffic signal control device based on 3D laser radar proposed by the present invention on the cantilever signal light pole at the urban intersection, as shown in FIG. 2 . The 3D laser radar detection unit in the device is responsible for real-time and dynamic detection, identification, tracking and information extraction of the traffic flow in the direction of the intersection entrance and the pedestrian flow in the crosswalk; Accurate, multi-resolution traffic flow information, optimization of traffic signal phase, phase sequence, and green light duration in the direction of entrance, information interaction, signal coordination, dynamic monitoring and safety warning of pedestrian flow in crosswalks; traffic signal drive control unit according to traffic signal The traffic signal scheme formed by the interactive coordination unit drives the traffic signal display module, traffic signal prompt module, and traffic safety early warning module, and controls the lighting time, lighting time, lighting status, and display pattern of the traffic signal light group and information display screen , to display information and safety warnings.

A self-perception interactive traffic signal control device based on 3D laser radar proposed by the present invention, its features mainly include:

1) 3D lidar detection unit

The 3D lidar detection unit is composed of a 3D lidar sensor and a multi-core laser point cloud micro-processing module. The 3D lidar sensor and the multi-core laser point cloud micro-processing module are connected and transmitted by EMIF (External Memory Interface) bus; The sensor is used to scan and detect the traffic flow in the direction of the intersection entrance and the pedestrian flow in the crosswalk to form a laser point cloud data frame, and transmit the laser point cloud data frame to the multi-core laser point cloud micro-processing module through the EMIF bus; the multi-core laser point cloud micro-processing The module is responsible for data filtering and information extraction of the laser point cloud data frame transmitted by the 3D lidar sensor, and extracts the traffic flow parameter information of the vehicle flow and the flow of people from the laser point cloud data frame.

2) Traffic signal interaction coordination unit

The traffic signal interaction coordination unit is composed of communication module, FPGA main control module, DSP safety monitoring module, and PCB backplane. The communication module, FPGA main control module, and DSP signal regulation module use EMIF bus for connection and data communication transmission. The layout is fixed on the board; at the same time, the traffic signal interaction coordination unit is connected and data communication transmission with the 3D lidar detection unit and the traffic signal drive control unit through the EMIF bus;

The communication module in the traffic signal interaction coordination unit is responsible for information interaction and sharing with the self-perception interactive traffic signal control devices installed on other cantilever light poles in the same intersection; the FPGA main control module is based on the traffic flow provided by the lidar detection unit. Parameter information, optimize the traffic signal phase and green light duration for the corresponding entrance direction, and perform phase, phase sequence, and green light duration optimization through the communication module and the self-sensing interactive traffic signal control device on other cantilever light poles at the intersection Coordination; the DSP safety monitoring module conducts dynamic monitoring and safety warning of the safety status of traffic flow and people flow in the direction of the intersection entrance, and automatically intervenes and quickly adjusts traffic signals in emergencies.

3) Traffic signal drive control unit

Traffic signal drive control unit consists of traffic signal display module, traffic signal prompt module, traffic safety early warning module, signal light group, LED micro display screen, 6U VPX switch signal interface board, traffic signal display module, traffic signal prompt module, traffic safety early warning The module, the signal light group, and the LED micro-display screen are connected and communicated with the 6U VPX switch signal interface board through the serial port; the traffic signal display module is responsible for the traffic signal control scheme transmitted from the traffic signal interaction coordination unit, driving and controlling the signal light group Turn-on time, turn-on state, turn-on pattern; the traffic signal prompt module controls the information prompt content of the LED micro-display screen according to the traffic signal control scheme transmitted by the traffic signal interaction coordination unit, and the traffic status and time of the flow of people and vehicles Carry out information prompts and dynamic guidance; the traffic safety warning module provides safety warnings for the flow of people and vehicles in the direction of the intersection entrance according to the traffic safety warning instructions sent by the traffic signal interaction coordination unit.

Description of drawings

Figure 1: Functional structure diagram of a self-aware interactive traffic signal control device based on 3D lidar;

Figure 2: Installation and layout diagram of a self-sensing interactive traffic signal control device based on 3D lidar;

Figure 3: Schematic diagram of the vehicle detection space drawn by the multi-core laser point cloud microprocessing module;

Figure 4: Schematic diagram of the pedestrian detection space drawn by the multi-core laser point cloud microprocessing module.

Detailed ways

A self-sensing interactive traffic signal control device based on 3D laser radar according to the present invention, the device is mainly composed of a 3D laser radar detection unit, a traffic signal interaction coordination unit, and a traffic signal drive control unit, as shown in Figure 1 and Figure 2 As shown; the device uses a 3D lidar sensor to realize the independent perception of traffic flow and passenger flow in the direction of the intersection entrance and the real-time, high-precision, and multi-resolution automatic extraction of traffic flow parameters, and conduct refined traffic signal optimization, coordination, and control. Improve the traffic efficiency of intersections and improve the utilization rate of road traffic resources.

A kind of self-sensing interactive traffic signal control device based on 3D laser radar proposed by the present invention, the specific process of its work is:

1) 3D lidar detection unit

The 3D laser radar detection unit is composed of a laser radar sensor and a multi-core laser point cloud micro-processing module. The 3D laser radar sensor and the multi-core laser point cloud micro-processing module are connected and transmitted by EMIF (External Memory Interface) bus; the laser radar sensor uses It is used to scan and detect the traffic flow in the direction of the entrance of the intersection and the flow of people in the crosswalk to form a laser point cloud data frame, and transmit the laser point cloud data frame to the multi-core laser point cloud micro-processing module through the EMIF bus; the multi-core laser point cloud micro-processing module is responsible for Data filtering and information extraction are performed on the laser point cloud data frame transmitted by the 3D lidar sensor, and the traffic flow parameter information of the vehicle flow and the flow of people is extracted from the laser point cloud data frame. The specific working steps are as follows:

Step1: The 3D laser radar sensor emits laser beams at a certain frequency and rotates the laser refracting mirror, and realizes 3D scanning of the road traffic environment in the detection direction by receiving laser reflected beams, and forms 3D laser point cloud images in the form of laser point clouds , when the 3D lidar sensor completes a 3D scan of the road traffic environment within the detection space, a 3D laser point cloud data frame is formed, and the data frame contains the three-dimensional coordinate information (X, Y, Z coordinates) of the laser point cloud , laser intensity, laser ID, laser horizontal rotation angle, laser vertical angle, laser distance, time stamp;

Step2: The multi-core laser point cloud microprocessing module divides the scanning space of the 3D lidar sensor into two subspaces, one is the vehicle detection space VEH_Ω in the direction of entrance, and the other is the pedestrian detection space PED_Ω of the crosswalk in the detection direction, as shown in Figure 3 and Figure 4 ; Among them: VEH_Ω is a cuboid of VL×VW×H, PED_Ω is a cuboid of PL×PW×H, VL is the distance between the farthest position in the direction of the entrance detected by the 3D lidar sensor and the parking line in the direction of the entrance; VW is 3D The width of the entrance road in the detection direction of the laser radar sensor; H is the height of the 3D laser radar sensor from the ground; PL is the length of the pedestrian crossing in the detection direction of the 3D laser radar sensor; PW is the width of the pedestrian crossing in the detection direction of the 3D laser radar sensor;

Step3: The multi-core laser point cloud microprocessing module extracts the laser point cloud in the space range of VEH_Ω and PED_Ω respectively from the laser point cloud data frame, and constructs a three-dimensional space matrix V(VEH_Ω) of MV×NV×KV, v _ijk ∈ The three-dimensional space matrix P(PED_Ω) of V(VEH_Ω) and MP×NP×KP, p _ijk ∈ P(PED_Ω);

In VEH_Ω: v _ijk is the i-th row of the X coordinate axis, the j-th column of the Y coordinate axis, and the laser intensity of the k-th layer of the Z coordinate axis in VEH_Ω; i=1, 2,..., MV; j=1, 2 ,...,NV; k=1,2,...,KV; MV is the total number of rows of the X-axis laser point cloud in VEH_Ω; NV is the total number of columns of the Y-axis laser point cloud in VEH_Ω; KV is the Z coordinate in VEH_Ω The total number of layers of the axis laser point cloud;

In PED_Ω: p _ijk is the i-th row of the X coordinate axis, the j-th column of the Y coordinate axis, and the laser intensity of the k-th layer of the Z coordinate axis in PED_Ω; i=1, 2,..., MP; j=1, 2 ,…,NP; k=1,2,…,KP; MP is the total number of rows of the X-axis laser point cloud in PED_Ω; NV is the total number of columns of the Y-axis laser point cloud in PED_Ω; KP is the Z coordinate in PED_Ω The total number of layers of the axis laser point cloud;

In VEH_Ω and PED_Ω, the direction of the intersection entrance scanned by the 3D lidar sensor is the direction of the Y coordinate axis; the direction of the pedestrian crossing in the direction of the intersection entrance scanned by the 3D lidar sensor is the direction of the X coordinate axis; the vertical direction of the intersection ground is the direction of the Z coordinate axis;

Step4: Perform background filtering on the laser point cloud within the VEH_Ω space range:

分别为长方体VEH_Ω中的上层、下层第i行第j列的长方体A；

分别为长方体VEH_Ω中的左壁、右壁第j列第k层的长方体A；i＝1，2，…，MV′；j＝1，2，…，NV′；k＝1，2，…，KV′；

令δ(A)为长方体A的激光点云密度，计算长方体A的激光点云密度波动率：Step4.1: Define A as a cuboid that can contain UV×VV×WV laser points, A∈VEH_Ω,

Respectively, cuboid A in the upper and lower layers of the i-th row and j-th column in the cuboid VEH_Ω;

Respectively cuboid A in the left and right walls of the cuboid VEH_Ω in the jth column and the kth layer; i=1, 2,..., MV'; j=1, 2,..., NV'; k=1, 2,... , KV′;

Let δ(A) be the laser point cloud density of cuboid A, and calculate the laser point cloud density fluctuation rate of cuboid A:

分别为以A为截面的长方体VEH_Ω中上层、下层、左壁、右壁的激光点云平均密度；

分别为长方体VEH_Ω中上层、下层第i行第j列的长方体A的激光点云密度波动率；

分别为长方体VEH_Ω中左壁、右壁第j列第k层的长方体A的激光点云密度波动率；in:

Respectively, the average density of the laser point cloud in the upper layer, lower layer, left wall, and right wall of the cuboid VEH_Ω with A as the cross-section;

Respectively, the laser point cloud density fluctuation rate of the cuboid A in the upper and lower layers of the cuboid VEH_Ω in the i-th row and j-th column;

Respectively, the laser point cloud density fluctuation rate of the cuboid A in the jth column and the kth layer of the left wall and the right wall in the cuboid VEH_Ω;

Step4.2: Determine whether the cuboid A in the upper layer, lower layer, left wall, and right wall of the cuboid VEH_Ω is the background laser point cloud image, and filter the background:

则判断长方体VEH_Ω中上层第i行第j列的长方体A所包含的激光点云为背景激光点云；like

Then it is judged that the laser point cloud contained in the cuboid A in the i-th row and j-column of the upper layer in the cuboid VEH_Ω is the background laser point cloud;

则判断长方体VEH_Ω中下层第i行第j列的长方体A所包含的激光点云为背景激光点云；like

Then it is judged that the laser point cloud contained in the cuboid A contained in the i-th row and j-column in the lower layer of the cuboid VEH_Ω is the background laser point cloud;

则长方体VEH_Ω中左壁第j列第k层的长方体A所包含的激光点云为背景激光点云；like

Then the laser point cloud contained in the cuboid A in the jth column on the left wall of the kth layer in the cuboid VEH_Ω is the background laser point cloud;

则长方体VEH_Ω中右壁第j列第k层的长方体A所包含的激光点云为背景激光点云；like

Then the laser point cloud contained in the cuboid A in the jth column and the kth layer on the right wall of the cuboid VEH_Ω is the background laser point cloud;

Among them: τ ^UP (T), τ ^DW (T), τ ^LF (T), τ ^RT (T) are respectively the background laser light of the cuboid A in the upper layer, lower layer, left wall, and right wall of the cuboid VEH_Ω in the T judgment period Point cloud density fluctuation critical threshold;

Step4.3: Repeat Step4.1 to Step4.2 to judge the background of all cuboids A in the upper layer, lower layer, left wall, and right wall of the cuboid VEH_Ω, and remove all background laser point clouds to form a new three-dimensional space matrix V '(VEH_Ω), v' _ijk ∈ V'(VEH_Ω);

Step5: Perform background filtering on the laser point cloud within the PED_Ω space range:

分别为长方体PED_Ω中的上层、下层第i行第j列的长方体B；

分别为长方体PED_Ω中的左壁、右壁第j列第k层的长方体B；i＝1，2，…，MP′；j＝1，2，…，NP′；k＝1，2，…，KP′；

令δ(B)为长方体B的激光点云密度，计算长方体B的激光点云密度波动率：Step5.1: Define B as a cuboid that can contain UP×VP×WP laser points, B∈PED_Ω,

Respectively, cuboid B in the upper and lower layers of the i-th row and j-th column in the cuboid PED_Ω;

Respectively, cuboid B in the left and right walls of the cuboid PED_Ω in the jth column and the kth layer; i=1, 2,..., MP'; j=1, 2,..., NP'; k=1, 2,... , KP′;

Let δ(B) be the laser point cloud density of cuboid B, and calculate the laser point cloud density fluctuation rate of cuboid B:

分别为以B为截面的长方体PED_Ω中上层、下层、左壁、右壁的激光点云平均密度；

分别为长方体PED_Ω中上层、下层第i行第j列的长方体B的激光点云密度波动率；

分别为长方体PED_Ω中左壁、右壁第j列第k层的长方体B的激光点云密度波动率；in:

Respectively, the average density of the laser point cloud in the upper layer, lower layer, left wall, and right wall of the cuboid PED_Ω with B as the cross-section;

Respectively, the laser point cloud density fluctuation rate of the cuboid B in the upper and lower layers of the cuboid PED_Ω at row i and column j;

Respectively, the laser point cloud density fluctuation rate of the cuboid B in the jth column and the kth layer of the left wall and right wall in the cuboid PED_Ω;

Step5.2: Determine whether the cuboid B in the upper layer, lower layer, left wall, and right wall of the cuboid PED_Ω is a background image, and filter the background:

则判断长方体PED_Ω中上层第i行第j列的长方体B所包含的激光点云为背景激光点云；like

Then it is judged that the laser point cloud contained in the cuboid B in the upper i-th row and j-th column of the cuboid PED_Ω is the background laser point cloud;

则判断长方体PED_Ω中下层第i行第j列的长方体B所包含的激光点云为背景激光点云；like

Then it is judged that the laser point cloud contained in the cuboid B contained in the i-th row and j-th column in the lower layer of the cuboid PED_Ω is the background laser point cloud;

则长方体PED_Ω中左壁第j列第k层的长方体B所包含的激光点云为背景激光点云；like

Then the laser point cloud contained in the cuboid B in the jth column on the left wall of the kth layer in the cuboid PED_Ω is the background laser point cloud;

则长方体PED_Ω中右壁第j列第k层的长方体B所包含的激光点云为背景激光点云；like

Then the laser point cloud contained in the cuboid B in the jth column and the kth layer on the right wall of the cuboid PED_Ω is the background laser point cloud;

Among them: λ ^UP (T), λ ^DW (T), λ ^LF (T), λ ^RT (T) are respectively the background laser light of the cuboid B in the upper layer, lower layer, left wall, and right wall of the cuboid PED_Ω in the Tth judgment cycle Point cloud density fluctuation critical threshold;

Step5.3: Repeat Step5.1 to Step5.2 to judge the background of all cuboids B in the upper layer, lower layer, left wall, and right wall of the cuboid PED_Ω, and remove all background laser point clouds to form a new three-dimensional space matrix P '(PEH_Ω), p' _ijk ∈ P'(PED_Ω);

Step6: Divide the X and Y axis planes of VEH_Ω into NC×NL grids, and each grid is a cuboid E of WL×LV×H; where: WL is the lane width in the direction of the entrance of the detection direction of the 3D lidar sensor ; LV is the average length of the vehicle; H is the height of the 3D lidar sensor from the ground; E ∈ VEH_Ω, E _ij is the cuboid in which VEH_Ω is the i-th row and the j-th lane; where: i=1, 2,..., NC; j = 1, 2, ..., NL;

Step6.1: Calculate the laser point cloud density δ(E) of the cuboid E in VEH_Ω

为E中所有激光点云密度之和；

为长方体E的体积；in:

is the sum of all laser point cloud densities in E;

is the volume of cuboid E;

Step6.2: Determine the existence of vehicles in the cuboid E:

(1) If δ(E) _≥∈V , it is judged that there is a vehicle in the cuboid E;

和

若

且

则车辆存在于长方体E的前部，将

和

合并，标记为新的长方体E′；若

且

则车辆存在于长方体E的后部，将

和

合并，标记为新的长方体E′；(2) If ∈ _bV ≤ δ(E)<∈ _V , split the cuboid E into two parts with equal volume

and

like

and

Then the vehicle exists in the front of the cuboid E, and the

and

Merged, marked as a new cuboid E′; if

and

Then the vehicle exists at the rear of the cuboid E, and the

and

Merge, marked as a new cuboid E';

为长方体E相邻的前一个长方体E的后半部分；

为长方体E相邻的后一个长方体E的前半部分；∈_V为长方体E车辆存在的临界密度；∈_bV为长方体E车辆占用的临界密度；in:

is the second half of the previous cuboid E adjacent to cuboid E;

is the first half of the next cuboid E adjacent to cuboid E; ∈ _V is the critical density of vehicles in cuboid E; ∈ _bV is the critical density occupied by cuboid E vehicles;

Step6.3: The minimum number of adjacent points of a given vehicle is Min_VehPts, and the neighborhood radius is R_Veh; traverse all the laser points Li_Pt in the cuboid E or E′, and find out the number of laser points NumPt(Li_Pt) within the radius R_Veh of each laser point Li_Pt ); if NumPt(Li_Pt)≥Min_VehPts, then mark the laser point Li_Pt as the vehicle center of mass VehCore_Pt in the cuboid E or E′;

Step6.4: Repeat Step6.1 to Step6.3 to obtain the number of vehicles VehNum and queue length VehQue of each entrance lane in the current laser point cloud data frame VEH_Ω according to the position information and quantity of the vehicle centroid points;

Step7: Divide the X and Y axis planes of PED_Ω into NP×NT grids, and each grid is a cuboid S of WP×PT×H; where: WP is the average longitudinal space distance required for pedestrians to walk; PT H is the height of the lidar sensor from the inside; S ∈ PED_Ω, _Sij is PED_Ω is the cuboid S in row i and column j in the pedestrian crossing; where: i=1, 2,..., NP; j = 1, 2, ..., NT;

Step7.1: Calculate the laser point cloud density δ(S) of the cuboid S in PED_Ω

为S中所有激光点云密度之和；

为长方体S的体积；in:

is the sum of all laser point cloud densities in S;

is the volume of cuboid S;

Step7.2: Determine the existence of pedestrians in the cuboid S:

(1) If δ(S) _≥∈p , it is judged that there are pedestrians in the cuboid S;

和

或左右两部分

和

若

且

则行人存在于长方体S的前部，将

和

合并，标记为新的长方体S′；若

且

则车辆存在于长方体S的后部，将

和

合并，标记为新的长方体S′；若

且

则行人存在于长方体S的右侧，将

和

合并，标记为新的长方体S′；若

且

则行人存在于长方体S的左侧，将

和

合并，标记为新的长方体S′(2) If ∈ _bp ≤ δ(S)<∈ _p , split the cuboid S into two parts with equal volume

and

or left and right parts

and

like

and

Then the pedestrian exists in the front of the cuboid S, and the

and

Merged, marked as a new cuboid S'; if

and

Then the vehicle exists at the rear of the cuboid S, and the

and

Merged, marked as a new cuboid S'; if

and

Then the pedestrian exists on the right side of the cuboid S, and the

and

Merged, marked as a new cuboid S'; if

and

Then the pedestrian exists on the left side of the cuboid S, and the

and

Merged, labeled as new cuboid S'

为长方体S相邻的前一个长方体S的后半部分；

为长方体S相邻的后一个长方体S的前半部分；∈_p为长方体S行人存在的临界密度；∈b_p为长方体行人占用的临界密度；in:

is the second half of the previous cuboid S adjacent to cuboid S;

is the first half of the next cuboid S adjacent to cuboid S; ∈ _p is the critical density of pedestrians in cuboid S; ∈ b _p is the critical density of pedestrians in cuboid;

Step7.3: The minimum number of adjacent points of a given pedestrian is Min_PedPts, and the neighborhood radius is R_Ped; traverse all the laser points Li_Pt in the cuboid S or S′, and find out the number of laser points NumPt(Li_Pt) within the neighborhood radius R_Ped of each laser point Li_Pt ); if NumPt(Li_Pt)≥Min_PedPts, then mark the laser point Li_Pt as the pedestrian centroid PedCore_Pt in the cuboid S or S′;

Step7.4: Repeat Step7.1 to Step7.3, according to the position information and number of pedestrian centroid points, obtain the number of pedestrians VPedNum and pedestrian location information PedLoInfo in the pedestrian crossing and pedestrian crossing waiting area in the current laser point cloud data frame PED_Ω ;

Step8: Repeat the processing of Step2 to Step7 for each frame of data transmitted by the 3D lidar sensor, track the position of vehicles and pedestrians in each frame, obtain the running trajectories VedTrace and PedTrace of vehicles and pedestrians, and pass vehicles and pedestrians Changes in the position of pedestrians, obtain the running speed information VehSpeedInfo and PedSpeedInfo of vehicles and pedestrians;

Step9: The multi-core laser point cloud micro-processing module transmits the number, position, speed, and running track information of vehicles and pedestrians in the detection area to the traffic signal interaction coordination unit through the EMIF bus.

2) Traffic signal interaction coordination unit

The traffic signal interaction coordination unit is composed of communication module, FPGA main control module, DSP safety monitoring module, and PCB backplane. The communication module, FPGA main control module, and DSP signal regulation module use EMIF bus for data communication transmission, and the The layout is fixed; the communication module is responsible for information interaction and sharing with the self-sensing interactive traffic signal control devices installed on other cantilever light poles in the same intersection; the FPGA main control module is based on the traffic flow parameter information provided by the lidar detection unit, Optimize the traffic signal phase and green light duration for the corresponding entrance direction, and coordinate the phase, phase sequence, and green light duration with the self-sensing interactive traffic signal control device on other cantilever light poles at the intersection through the communication module; DSP The safety monitoring module conducts dynamic monitoring and safety warnings on the safety status of traffic flow and pedestrian flow in crosswalks, and automatically intervenes and quickly adjusts traffic signals in emergencies; the specific working steps are as follows:

Step1: The FPGA main control module predicts the queuing traffic flow of each entrance lane in the detection direction at a certain time interval period T_GAP:

Among them: Vol(t+T_GAP), Vol(t), and Vol(t-T_GAP) are respectively the flow of the intersection entrance road at the next forecast period interval T_GAP, the current forecast period, and the previous forecast period interval T_GAP at time t; Vel(t) and Vel(t-T_GAP) are the current forecast period at time t and the average vehicle speed at the intersection entrance of the last forecast period interval T_GAP; α and β are correction parameters respectively;

Step2: The FPGA main control module estimates the time required for the queuing vehicles in each entrance lane to dissipate:

为排队车辆平均消散速度；Among them: disT is the dissipation time of queuing vehicles in the lane; μ is the start delay of queuing vehicles; γ is the correction coefficient of queuing dissipation time;

is the average dissipation speed of queuing vehicles;

Step3: The FPGA main control module determines the queuing vehicle dissipation time required for each direction of the entrance lane:

disT_ST=max(disT_ST ₁ , disT_ST ₂ , . . . , disT_ST _n )

disT_LT=max(disT_LT ₁ , disT_LT ₂ , . . . , disT_LT _m )

Among them: _{disT_ST} , disT_LT are the queuing vehicle dissipation time _required for the straight-going direction and left-turn direction of _the entrance lane respectively; The queuing vehicle dissipation time required for the lane; disT_LT ₁ , disT_LT ₂ , ..., disT_LT _m are respectively the queuing vehicle dissipation time required for the first, 2, ..., m left-turn lanes in the entrance lane;

Step4: The FPGA main control module calculates the time difference between the dissipation time disT_ST of the queuing vehicles in the through direction of the entrance lane and disT_LT of the queuing vehicles in the left turn direction:

ΔdisT=|disT_ST-disT_LT|

(1) If ΔdisT≤ξ, the main control module of the FPGA merges the phases of the through traffic signal and the left turn signal phase of the entrance lane into the same signal phase phase, and the green light time of the phase phase is max(disT_ST, disT_LT);

(2) If ΔdisT>ξ, the FPGA main control module obtains the dissipation times disT_ST' and disT_LT' of the queuing vehicles in the through direction and left turn direction of the opposite entrance lane through the communication module;

①If |disT_ST-disT_ST'|≤ξ, the FPGA main control module coordinates with the opposite FPGA main control module through the communication module, and merges the phase of the through traffic signal of the entrance lane and the phase of the through traffic signal of the opposite entrance lane into the same A signal phase phase, the green light time of the phase phase is max(disT_ST, disT_ST');

②If |disT_LT-disT_LT'|≤ξ, the FPGA main control module coordinates with the opposite FPGA main control module through the communication module, and merges the left-turn traffic signal phase of the entrance lane with the left-turn traffic signal phase of the opposite entrance lane For the same signal phase phase, the green light time of the phase phase is max(disT_LT, disT_LT');

③If |disT_ST-disT_ST’|≤ξ and |disT_LT-disT_LT’|>ξ, the FPGA main control module coordinates with the opposite FPGA main control module through the communication module:

The phase of the through traffic signal of the entrance lane and the phase of the through traffic signal of the opposite entrance lane are combined into the same signal phase phase, and the green light time of the phase phase is max(disT_ST, disT_ST’);

If disT_LT>disT_LT', and maxVOL_ST(t+T_GAP)>maxVOL_LT'(t+T_GAP), then the left-turn signal phase of the entrance lane and the left-turn signal phase of the opposite entrance lane are combined into the same signal phase phase , the green light time in the phase phase is min(disT_LT, disT_LT'); and a superimposed phase phase is added to the entrance lane, which allows the through direction and left-turn traffic direction of the entrance direction to pass at the same time, and the green light time in the superimposed phase phase is |disT_LT- disT_LT'|;

If disT_LT>disT_LT', and maxVOL_ST(t+T_GAP)<maxVOL_LT'(t+T_GAP), the phase of the left-turn traffic signal of the entrance lane and the phase of the left-turn traffic signal of the opposite entrance lane are merged into the same signal phase phase , the green light time of the phase phase is disT_LT;

If disT_LT<disT_LT', and maxVOL_LT(t+T_GAP)>maxVOL_ST'(t+T_GAP), the phase of the left-turn traffic signal of the entrance lane and the phase of the left-turn traffic signal of the opposite entrance lane are combined into the same signal phase phase , the green light time of the phase phase is disT_LT';

If disT_LT<disT_LT', and maxVOL_LT(t+T_GAP)<maxVOL_ST'(t+T_GAP), the phase of the left-turn traffic signal of the entrance lane and the phase of the left-turn traffic signal of the opposite entrance lane are merged into the same signal phase phase , the green light time in the phase phase is min(disT_LT, disT_LT'); and add a superimposed phase phase to the opposite entrance lane, allowing the opposite entrance to pass in the through direction and the left turn direction at the same time, and the green light time in the superimposed phase phase for |disT_LT-disT_LT'|;

④If |disT_ST-disT_ST’|>ξ and |disT_LT-disT_LT’|≤ξ, the FPGA main control module coordinates with the opposite FPGA main control module through the communication module:

Merge the phase of the left-turn traffic signal of the entrance lane and the phase of the left-turn traffic signal of the opposite entrance lane into the same signal phase phase, and the green light time of the phase phase is max(disT_LT, disT_LT’);

If disT_ST>disT_ST', and maxVOL_LT(t+T_GAP)>maxVOL_ST'(t+T_GAP), the phase of the through signal of the entrance lane and the phase of the through signal of the opposite entrance lane are combined into the same signal phase phase, the phase The green light time of the phase is min(disT_ST, disT_ST'); and a superimposed phase phase is added to the entrance lane, which allows the through direction and left-turn traffic direction of the entrance direction to pass at the same time, and the green light time of the superimposed phase phase is |disT_ST-disT_ST' |;

If disT_ST>disT_ST', and maxVOL_LT(t+T_GAP)<maxVOL_ST'(t+T_GAP), then the phase of the through traffic signal of the entrance lane and the phase of the through traffic signal of the opposite entrance lane are combined into the same signal phase phase, the phase The green light time of the stage is disT_ST;

If disT_ST<disT_ST', and maxVOL_ST(t+T_GAP)>maxVOL_LT'(t+T_GAP), the phase of the left-turn traffic signal of the entrance lane and the phase of the left-turn traffic signal of the opposite entrance lane are merged into the same signal phase phase , the green light time of the phase phase is disT_LT';

If disT_ST<disT_ST', and maxVOL_ST(t+T_GAP)<maxVOL_LT'(t+T_GAP), the phase of the through traffic signal of the entrance lane and the phase of the through traffic signal of the opposite entrance lane are merged into the same signal phase phase, the phase The green light time of the stage is min(disT_ST, disT_ST'); and a superimposed phase phase is added to the opposite entrance lane, which allows the passage in the through direction and the left turn direction of the entrance direction to pass at the same time, and the green light time of the superimposed phase phase is |disT_ST- disT_ST'|;

Among them: ξ phase stage splitting critical threshold; maxVOL_ST(t+T_GAP), maxVOL_ST′(t+T_GAP) are respectively the maximum through lanes in the through direction of the entrance lane and the through direction of the opposite entrance lane in the next prediction interval T_GAP at time t Flow; maxVOL_LT(t+T_GAP), maxVOL_LT′(t+T_GAP) are respectively the maximum left-turn lane traffic in the left-turn direction of the entrance lane and the left-turn direction of the opposite entrance lane in the next prediction interval T_GAP at time t;

Step5: The FPGA main control module communicates with other FPGA main control modules at the intersection through the communication module. According to the green light duration of each passing phase in each entrance direction calculated by each FPGA main control module, according to the Webster intersection delay calculation formula, adopt The principle of minimum delay at intersections is to coordinate and determine the execution sequence of each signal phase, that is, the phase sequence;

Step6: The DSP safety monitoring module monitors the running conditions of vehicles and pedestrians in real time:

Step6.1: When the green light of the vehicle signal is about to end, the DSP safety monitoring module monitors the vehicles that are about to enter the intersection in the direction of the entrance in real time, and calculates whether the current vehicle can safely stop or pass through the intersection:

Among them: X ₀ is the minimum safe parking distance that the vehicle can decelerate and stop; X _C is the maximum safe distance that the vehicle can accelerate to pass; V ₀ is the running speed of the vehicle that is about to enter the intersection; σ is the average braking reaction time of the driver; α _- is the average braking deceleration of the vehicle; α ₊ is the average braking acceleration of the vehicle; greenT is the remaining transit time; w is the intersection width; l is the average vehicle length;

Step6.2: If X ₀ >X _C , the vehicle can neither pass through nor stop safely, then the DSP safety monitoring module coordinates with the FPGA main control module to prolong the green light time of the vehicle running direction so that it can pass safely and The traffic signal drive control module sends a DilemmaType1 instruction packet;

Step6.3: If X ₀ <X _C , then the DSP safety monitoring module coordinates with the traffic signal drive control unit to flash the signal light in the direction of the vehicle, prompting the driver to slow down and stop or speed up to pass, and sends a DilemmaType2 command to the traffic signal drive control module data pack;

Step6.4: The DSP safety monitoring module monitors the running trajectories of pedestrians in the crosswalk and vehicles in the direction of the entrance in real time. Once a conflict between the running trajectories of pedestrians and vehicles is detected, the DSP safety monitoring module sends a signal to the traffic signal drive control unit. "Danger, please pay attention to safety!" security alert DangerInfo command packet;

Step6.5: When the green light of the vehicle signal is about to end, the DSP safety monitoring module monitors whether there are pedestrians in the crosswalk in real time. Once a pedestrian is detected at the border of the crosswalk at the edge of the road, the DSP safety monitoring module sends "wait patiently" to the traffic signal drive control unit , please return to the waiting area!" security warning PedWarningType1 instruction packet;

Step6.6: During the red light of the vehicle signal and the green light of the pedestrian signal, the DSP safety detection module monitors whether the pedestrian is walking on the crosswalk in real time. Once it detects that the track of a pedestrian walking is not on the crosswalk, the DSP safety monitoring module drives the control unit to the traffic signal Send the "civilized traffic, please walk the zebra crossing!" safety warning PedWarningType2 instruction data packet;

Step7: The FPGA main control module and DSP safety detection module transmit the traffic signal control scheme, including the green light time of each signal phase, the execution sequence of the signal phase, and the traffic safety warning command, to the traffic signal drive control unit through the EMIF bus;

3) Traffic signal drive control unit

Traffic signal drive control unit consists of traffic signal display module, traffic signal prompt module, traffic safety early warning module, signal light group, LED micro display screen, 6U VPX switch signal interface board, traffic signal display module, traffic signal prompt module, traffic safety early warning The module, the signal lamp group, and the LED micro-display screen are connected and communicated with the 6U VPX switch signal interface board through the serial port; the traffic signal display module is responsible for the traffic signal control scheme transmitted from the traffic signal coordination unit, and drives and controls the activation of the signal lamp group. Lighting time, lighting state, and lighting pattern; the traffic signal prompt module controls the information prompt content of the LED micro-display screen according to the traffic signal control scheme transmitted by the traffic signal interaction coordination unit, and monitors the passing status and passing time of the flow of people and vehicles. Information prompts and dynamic guidance; the traffic safety early warning module provides safety warnings for the flow of people and vehicles at the intersection according to the traffic safety early warning instructions sent by the traffic signal interaction coordination unit.

Step1: The traffic signal display module receives the traffic signal control scheme sent by the traffic signal interaction coordination unit, converts the digital information of the traffic signal control scheme from the analog signal, drives and controls the lighting time, lighting color, and lighting of the signal light group sequence, the transition sequence and transition form of different light colors, and the graphics displayed by light colors;

Step2: The traffic signal prompt module receives the traffic signal control scheme sent by the traffic signal interaction coordination unit, drives and controls the LED micro-display screen, and prompts and guides the passing status, passing time, remaining time and traveling direction of the flow of people and vehicles;

Step3: The traffic safety warning module receives the traffic safety warning instructions sent by the traffic signal interaction coordination unit, and performs safety warnings according to different signal instructions:

Step3.1: When the traffic safety warning module receives the DilemmaType1 instruction packet, it takes over the control of the traffic signal display module on the signal light group, and according to the content of the DilemmaType1 instruction, prolongs the duration of the green light in the designated traffic direction, and releases the signal light when the extended time ends Group control, return the control right of the signal light group to the traffic signal display module;

Step3.2: When the traffic safety warning module receives the DilemmaType2 instruction packet, it takes over the control of the traffic signal display module on the signal light group, and according to the content of the DilemmaType2 instruction, it performs blinking control on the signal lights in the designated direction of traffic. When the blinking control ends, release For the control of the signal light group, return the control right of the signal light group to the traffic signal display module;

Step3.3: When the traffic safety early warning module receives the DangerInfo instruction data packet, it takes over the control of the traffic signal display module on the signal light group, and according to the content of the DangerInfo instruction, starts the red light flashing warning for the designated traffic direction, and displays it through voice and LED micro display Screen alarm "Danger, please pay attention to safety!";

Step3.4: When the traffic safety warning module receives the PedWarningType1 instruction packet, it will alarm the pedestrians "wait patiently, please return to the waiting area!" through voice and LED micro-display screen;

Step3.5: When the traffic safety warning module receives the PedWarningType2 command data packet, it will send an alarm to pedestrians "civilized traffic, please walk the zebra crossing!" through voice and LED micro-display screen.

Claims (3)

1. The utility model provides a self-perception interactive traffic signal controlling means based on 3D lidar which characterized in that:

the 3D laser radar detection unit consists of a laser radar sensor and a multi-core laser point cloud micro-processing module, and the 3D laser radar sensor and the multi-core laser point cloud micro-processing module are connected and transmitted in data communication by adopting an EMIF bus; the laser radar sensor is used for scanning and detecting traffic flow in the inlet direction of the intersection and pedestrian flow of a pedestrian crossing to form a laser point cloud data frame, and transmitting the laser point cloud data frame to the multi-core laser point cloud micro-processing module through an EMIF bus; the multi-core laser point cloud microprocessing module is responsible for carrying out data filtering and information extraction on a laser point cloud data frame transmitted by the 3D laser radar sensor and extracting traffic flow and people flow parameter information from the laser point cloud data frame; the specific working steps are as follows:

step1: the method comprises the following steps that a 3D laser radar sensor emits laser beams and rotates a laser refraction mirror surface at a certain frequency, 3D scanning of the road traffic environment in the detection direction is achieved by receiving laser reflection beams, a 3D laser point cloud image is formed in a laser point cloud mode, when the 3D laser radar sensor completes 3D scanning of the road traffic environment in the detection space range every time, a 3D laser point cloud data frame is formed, and the data frame comprises three-dimensional coordinate information, laser intensity, laser ID, a laser horizontal rotation direction angle, a laser vertical direction included angle, a laser distance and a timestamp of the laser point cloud;

step2: the multi-core laser point cloud microprocessing module divides a scanning space of the 3D laser radar sensor into two subspaces, wherein one subspace is a vehicle detection space VEH _ omega in the direction of an entrance, and the other subspace is a pedestrian detection space PED _ omega in a pedestrian crosswalk in the direction of detection; wherein: VEH _ omega is a cuboid with VL multiplied by VW multiplied by H, PED _ omega is a cuboid with PL multiplied by PW multiplied by H, and VL is the distance between the farthest position of the 3D laser radar sensor in the inlet direction and the stop line in the inlet direction; VW is the width of an entrance way in the detection direction of the 3D laser radar sensor; h is the height of the 3D laser radar sensor from the ground; PL is the length of a crosswalk in the direction detected by the 3D laser radar sensor; PW is the width of a pedestrian crossing in the direction detected by the 3D laser radar sensor;

step3: the multi-core laser point cloud microprocessing module respectively extracts the data frames at VEH _ omega and PED _ omega from the laser point cloud data framesRespectively constructing a MV multiplied by NV multiplied by KV three-dimensional space matrix V (VEH _ omega), V by laser point cloud in a space range _ijk e.V (VEH _ OMEGA) and MP × NP × KP three-dimensional space matrix P (PED _ OMEGA), P _ijk ∈P(PED_Ω)；

In VEH _ Ω: v. of _ijk The laser intensity of the ith row of the X coordinate axis, the jth column of the Y coordinate axis and the kth layer of laser points of the Z coordinate axis in VEH _ omega is shown; i =1,2, \8230;, MV; j =1,2, \ 8230;, NV; k =1,2, \ 8230;, KV; MV is the total row number of X coordinate axis laser point clouds in VEH _ omega; NV is the total column number of Y coordinate axis laser point clouds in VEH _ omega; KV is the total number of layers of Z coordinate axis laser point clouds in VEH _ omega;

in PED _ Ω: p is a radical of _ijk The laser intensity of the laser spot on the ith row of the X coordinate axis, the jth column of the Y coordinate axis and the kth layer of the Z coordinate axis in PED _ omega is set; i =1,2, \ 8230;, MP; j =1,2, \ 8230;, NP; k =1,2, \ 8230;, KP; MP is the total line number of X coordinate axis laser point clouds in PED _ omega; NV is the total column number of Y coordinate axis laser point clouds in PED _ omega; KP is the total number of layers of Z coordinate axis laser point clouds in PED _ omega;

the intersection inlet direction scanned by the 3D laser radar sensor in VEH _ omega and PED _ omega is the Y coordinate axis direction; the pedestrian crossing direction in the intersection inlet direction scanned by the 3D laser radar sensor is the X coordinate axis direction; the vertical direction of the ground at the intersection is the direction of a Z coordinate axis;

step4: background filtering is carried out on the laser point cloud in the VEH _ omega space range:

step4.1: defining A as a cuboid containing UV VV WV laser spots, A ∈ VEH _ Ω,

the cuboids A are respectively an upper-layer row and a lower-layer jth row in the cuboid VEH _ omega;

the cuboids A are respectively the jth layer of the left wall and the jth layer of the right wall in the cuboid VEH _ omega; i =1,2, \8230;, MV'; j =1,2, \8230;, NV'; k =1,2, \ 8230;, KV';

let δ (a) be the laser point cloud density of the cuboid a, calculate the laser point cloud density fluctuation rate of the cuboid a:

wherein:

the laser point cloud average densities of the middle upper layer, the lower layer, the left wall and the right wall of the cuboid VEH _ omega with the A as the section are respectively obtained;

laser point cloud density fluctuation rates of the cuboid A on the ith row and the jth column of the middle upper layer and the lower layer of the cuboid VEH _ omega are respectively set;

laser point cloud density fluctuation rates of the cuboid A on the jth column and kth layer of the left wall and the right wall in the cuboid VEH _ omega are respectively set;

step4.2: judging whether the cuboid A in the upper layer, the lower layer, the left wall and the right wall of the cuboid VEH _ omega is a background laser point cloud image or not, and filtering the background:

if it is

Then, the cuboid A in the ith row and the jth column of the upper layer in the cuboid VEH _ omega is judged to containThe laser point cloud is background laser point cloud;

if it is

Judging that the laser point cloud contained in the cuboid A in the ith row and the jth column of the lower layer in the cuboid VEH _ omega is background laser point cloud;

if it is

Then the laser point cloud contained in the jth column and kth layer of the cuboid A on the left wall in the cuboid VEH _ omega is background laser point cloud;

if it is

Then the laser point cloud contained in the jth row and kth layer of the cuboid A on the right wall of the cuboid VEH _ omega is background laser point cloud;

wherein: tau is ^UP (T)、τ ^DW (T)、τ ^LF (T)、τ ^RT (T) respectively judging the critical threshold values of the point cloud density fluctuation of the cuboid A background laser in the middle upper layer, the lower layer, the left wall and the right wall of the cuboid VEH _ omega in the Tth judging period;

step4.3: repeating Step4.1 to Step4.2, carrying out background judgment on all cuboids A in the upper layer, the lower layer, the left wall and the right wall of the cuboid VEH _ omega, and eliminating all background laser point clouds to form a new three-dimensional space matrix V '(VEH _ omega), V' _ijk ∈V′(VEH_Ω)；

Step5: background filtering the laser point cloud in the PED _ omega space range:

step5.1: defining B as a cuboid containing UP VP WP laser points, B e PED omega,

the cuboids B are respectively an upper-layer row and a lower-layer jth column in the cuboid PED _ omega;

the cuboids B are respectively the jth layer of the left wall and the jth layer of the right wall in the cuboid PED _ omega; i =1,2, \ 8230;,MP′；j＝1，2，…，NP′；k＝1，2，…，KP′；

let δ (B) be the laser point cloud density of the cuboid B, calculate the laser point cloud density fluctuation rate of the cuboid B:

wherein:

the laser point cloud average densities of the middle upper layer, the lower layer, the left wall and the right wall of the cuboid PED _ omega with the section B as the section are respectively;

laser point cloud density fluctuation rates of cuboids B of the ith row and the jth column of the middle upper layer and the lower layer of the cuboid PED _ omega are respectively set;

laser point cloud density fluctuation rates of the cuboid B on the jth column k layer of the left wall and the right wall in the cuboid PED _ omega are respectively set;

step5.2: judging whether the cuboid B in the upper layer, the lower layer, the left wall and the right wall in the cuboid PED _ omega is a background image or not, and filtering the background:

if it is

Judging that the laser point cloud contained in the cuboid B in the ith row and the jth column of the upper layer in the cuboid PED _ omega is background laser point cloud;

if it is

Judging that the laser point cloud contained in the cuboid B in the ith row and the jth column of the lower layer in the cuboid PED _ omega is background laser point cloud;

if it is

Then the laser point cloud contained in the cuboid B of the jth column and kth layer on the left wall of the cuboid PED _ omega is background laser point cloud;

if it is

Then the laser point cloud contained in the cuboid B of the jth column and kth layer on the right wall of the cuboid PED _ omega is background laser point cloud;

wherein: lambda [ alpha ] ^UP (T)、λ ^DW (T)、λ ^LF (T)、λ ^RT (T) respectively judging the critical threshold values of the point cloud density fluctuation of the background laser points of the cuboid B in the upper layer, the lower layer, the left wall and the right wall in the cuboid PED _ omega in the Tth judging period;

step5.3: repeating Step5.1 to Step5.2, performing background judgment on all cuboids B in the upper layer, the lower layer, the left wall and the right wall of the cuboid PED _ omega, and removing all background laser point clouds to form a new three-dimensional space matrix P '(PEH _ omega), P' _ijk ∈P′(PED_Ω)；

Step6: dividing an X-axis plane and a Y-axis plane of VEH _ omega into NC multiplied by NL grids, wherein each grid is internally provided with a cuboid E of WL multiplied by LV multiplied by H; wherein: WL is the lane width of the 3D laser radar sensor in the direction of the inlet; LV is the average length of the vehicle; h is the height of the 3D laser radar sensor from the ground; e is equal to VEH _ omega, E _ij The vehicle is a cuboid with VEH _ omega as the jth lane of the ith row; wherein: i =1,2, \ 8230;, NC; j =1,2, \8230;, NL;

step6.1: calculating the laser point cloud density delta (E) of E of the cuboid in VEH _ omega

Wherein:

the sum of the point cloud densities of all the laser points in the E;

is the volume of cuboid E;

step6.2: judging the existence of the vehicle in the cuboid E:

(1) If delta (E) is equal to or greater than epsilon _V Judging that the vehicle exists in the cuboid E;

(2) If e is _bV ≤δ(E)＜∈ _V The cuboid E is divided into a front part and a rear part which have the same volume

And

if it is

And is

The vehicle is present in the front of the cuboid E and will

And

merging and marking as a new cuboid E'; if it is

And is

The vehicle is present at the rear of the rectangular parallelepiped E, will

And

merging and marking as a new cuboid E';

wherein:

the rear half part of the front cuboid E adjacent to the cuboid E;

the front half part of the next cuboid E adjacent to the cuboid E; e is the same as _V Critical density for the presence of cuboid E vehicles; e is the same as _bV Is the critical density occupied by the cuboid E vehicle;

step6.3: the minimum number of neighbors of a given vehicle is Min _ VehPTs, and the radius of the neighbors is R _ Veh; traversing all laser points Li _ Pt in the cuboid E or E' to find out the number NumPT (Li _ Pt) of the laser points in the neighborhood radius R _ Veh of each laser point Li _ Pt; if NumPt (Li _ Pt) is greater than or equal to Min _ VehPTs, marking the laser point Li _ Pt as the vehicle mass center VehCore _ Pt in the cuboid E or E';

step6.4: repeating Step6.1 to Step6.3, and acquiring the number VehNum of vehicles and the queuing length VehQue of each entrance way in the current laser point cloud data frame VEH _ omega according to the position information and the number of the vehicle center of mass points;

step7: dividing an X-axis plane and a Y-axis plane of PED _ omega into NP multiplied by NT grids, wherein each grid is internally provided with a cuboid S of WP multiplied by PT multiplied by H; wherein: WP is the average longitudinal space distance required by the pedestrian to walk; PT is the transverse space distance of the pedestrian; h is the height of the laser radar sensor from the inside; s belongs to PED omega, S _ij PED _ omega is a cuboid S in the ith row and the jth column in the pedestrian crossing; wherein: i =1,2, \8230;, NP; j =1,2,…，NT；

step7.1: calculating the laser point cloud density delta (S) of the rectangular S in PED _ omega

Wherein:

the sum of the point cloud densities of all the laser points in the S;

is the volume of the cuboid S;

step7.2: judging the existence of the pedestrian in the cuboid S:

(1) If delta (S) ≧ epsilon _p Judging that the pedestrian exists in the cuboid S;

(2) If e is _bp ≤δ(S)＜∈ _p The cuboid S is split into a front part and a rear part with equal volume

And

or left and right parts

And

if it is

And is

The pedestrian is present in the front of the rectangular parallelepiped S and will

And

merging and marking as a new cuboid S'; if it is

And is

The vehicle is present at the rear of the rectangular parallelepiped S, will

And

merging and marking as a new cuboid S'; if it is

And is

The pedestrian is present on the right side of the rectangular parallelepiped S and will

And

merging and marking as a new cuboid S'; if it is

And is

The pedestrian is present on the left side of the rectangular parallelepiped S and will be

And

combined and marked as new cuboid S'

Wherein:

the cuboid S is the rear half part of the adjacent front cuboid S;

the front half part of the next cuboid S adjacent to the cuboid S; e is the same as _p Is the critical density of a pedestrian existing in a cuboid S; e is a _bp Is the critical density occupied by the pedestrian in the cuboid S;

step7.3: the minimum number of adjacent points of a given pedestrian is Min _ PedPts, and the radius of the adjacent points is R _ Ped; traversing all laser points Li _ Pt in the cuboid S or S' to find out the number NumPt (Li _ Pt) of laser points in the neighborhood radius R _ Ped of each laser point Li _ Pt; if NumPt (Li _ Pt) is more than or equal to Min _ PedPts, marking the laser point Li _ Pt as a pedestrian centroid PedCore _ Pt in the cuboid S or S';

step7.4: repeating Step7.1 to Step7.3, and acquiring the pedestrian number VPedNum and the pedestrian position information PedLoInfo of the pedestrian crossing crosswalk and the pedestrian crossing waiting area in the current laser point cloud data frame PED _ omega according to the position information and the number of the pedestrian centroid points;

step8: repeatedly carrying out the processing from Step2 to Step7 on data of each frame transmitted by the 3D laser radar sensor, tracking the positions of the vehicles and the pedestrians in each frame, acquiring the running tracks VedTrace and PedTrace of the vehicles and the pedestrians, and acquiring running speed information VehSpeedInfo and PedSpeedInfo of the vehicles and the pedestrians through the position change of the vehicles and the pedestrians;

step9: the multi-core laser point cloud microprocessing module transmits the number, position, speed and running track information of vehicles and pedestrians in the detection area to the traffic signal interaction coordination unit through an EMIF bus.

2. The 3D lidar based self-sensing interactive traffic signal control apparatus of claim 1, wherein:

the traffic signal interaction coordination unit consists of a communication module, an FPGA main control module, a DSP safety monitoring module and a PCB backboard, wherein the communication module, the FPGA main control module and the DSP signal regulation and control module adopt an EMIF bus to carry out data communication transmission and carry out layout fixation on the PCB backboard; the communication module is responsible for carrying out information interaction and sharing with self-sensing interactive traffic signal control devices which are arranged on other cantilever type lamp poles in the same intersection; the FPGA main control module optimizes the phase of the traffic signal and the duration of the green light in the corresponding entrance direction according to the traffic flow parameter information provided by the laser radar detection unit, and coordinates the phase, the phase sequence and the duration of the green light with a self-sensing interactive traffic signal control device on other cantilever lamp poles at the intersection through the communication module; the DSP safety monitoring module carries out dynamic monitoring and safety early warning on the safety states of the traffic flow and the pedestrian flow of the pedestrian crossing, and carries out automatic intervention and quick adjustment on traffic signals under the emergency traffic incident; the specific working steps are as follows:

step1: the FPGA main control module predicts the queuing traffic flow of each entrance road in the detection direction at a certain time interval period T _ GAP:

wherein: vol (T + T _ GAP), vol (T) and Vol (T-T _ GAP) are respectively the flow of an intersection inlet channel of the next prediction period interval T _ GAP, the current prediction period and the previous prediction period interval T _ GAP at the time T; vel (T) and Vel (T-T _ GAP) are respectively the average speed of vehicles at the intersection entrance road of the current prediction period and the last prediction period interval T _ GAP at the time T; alpha and beta are correction parameters respectively;

step2: the FPGA main control module estimates the time required for dissipation of queued vehicles of each entrance lane:

wherein: disT is the lane in-line vehicle dissipation time; mu queueVehicle start-up delay; gamma is a queuing dispersion time correction coefficient;

average dissipation speed for the queued vehicles;

step3: the FPGA main control module determines the dissipation time of queued vehicles required by each passing direction of an entrance lane:

disT_ST＝max(disT_ST ₁ ，disT_ST ₂ ，…，disT_ST _n )

disT_LT＝max(disT_LT ₁ ，disT_LT ₂ ，…，disT_LT _m )

wherein: disT _ ST and disT _ LT are respectively the dissipation time of the queuing vehicles required by the straight-going traffic direction and the left-turning traffic direction of the entrance lane; disT _ ST ₁ ，disT_ST ₂ ，…，disT_ST _n Respectively 1 st, 2 nd, \ 8230, the dissipation time of queuing vehicles required by n straight lanes; disT _ LT ₁ ，disT_LT ₂ ，…，disT_LT _m Respectively 1 st, 2 nd, \ 8230;, the dissipation time of queuing vehicles required by m left-turn lanes;

step4: the FPGA main control module calculates the time difference between the dissipation time disT _ ST of the vehicles queued in the straight-going passing direction of the entrance lane and the dissipation time disT _ LT of the vehicles queued in the left-turning passing direction:

ΔdisT＝|disT_ST-disT_LT|

(1) If the delta disT is less than or equal to xi, the FPGA main control module combines the straight traffic signal phase and the left-turn traffic signal phase of the entrance lane into the same signal phase stage, and the green time of the phase stage is max (disT _ ST, disT _ LT);

(2) If the delta disT is larger than xi, the FPGA main control module acquires the dissipation time disT _ ST 'and disT _ LT' of the queued vehicles in the straight-going traffic direction and the left-turning traffic direction of the opposite-entrance lane through the communication module;

(1) if the | disT _ ST-disT _ ST '| is less than or equal to xi, the FPGA main control module coordinates with the opposite FPGA main control module through the communication module, combining the straight traffic signal phase of the entrance lane and the straight traffic signal phase of the opposite entrance lane into the same signal phase stage, wherein the green time of the phase stage is max (disT _ ST, disT _ ST');

(2) if the absolute value of the DIST _ LT-DIST _ LT 'is less than or equal to xi, the FPGA main control module coordinates with the opposite FPGA main control module through the communication module, the left-turn traffic signal phase of the entrance lane and the left-turn traffic signal phase of the opposite entrance lane are combined into the same signal phase stage, and the green light time of the phase stage is max (DIST _ LT, DIST _ LT');

(3) if | disT _ ST-disT _ ST '| is less than or equal to xi and | disT _ LT-disT _ LT' | is greater than xi, the FPGA main control module coordinates with the opposite FPGA main control module through the communication module:

combining the straight traffic signal phase of the entrance lane and the straight traffic signal phase of the opposite entrance lane into the same signal phase stage, wherein the green time of the phase stage is max (disT _ ST, disT _ ST');

if disT _ LT > disT _ LT ' and maxVOL _ ST (T + T _ GAP) > maxVOL _ LT ' (T + T _ GAP), merging the left-turn traffic signal phase of the entrance lane and the left-turn traffic signal phase of the opposite entrance lane into the same signal phase stage, wherein the green time of the phase stage is min (disT _ LT, disT _ LT '); a phase superposition stage is added in an entrance lane, the straight traffic direction and the left-turn traffic direction in the entrance direction are allowed to be released simultaneously, and the green light time in the phase superposition stage is | disT _ LT-disT _ LT' |;

if disT _ LT is greater than disT _ LT 'and maxVOL _ ST (T + T _ GAP) < maxVOL _ LT' (T + T _ GAP), merging the left-turning traffic signal phase of the entrance lane and the left-turning traffic signal phase of the opposite entrance lane into the same signal phase stage, wherein the green time of the phase stage is disT _ LT;

if disT _ LT is less than disT _ LT ' and maxVOL _ LT (T + T _ GAP) > maxVOL _ ST ' (T + T _ GAP), merging the left-turn traffic signal phase of the entrance lane and the left-turn traffic signal phase of the opposite entrance lane into the same signal phase stage, wherein the green time of the phase stage is disT _ LT ';

if disT _ LT is less than disT _ LT ' and maxVOL _ LT (T + T _ GAP) < maxVOL _ ST ' (T + T _ GAP), merging the left-turning traffic signal phase of the entrance lane and the left-turning traffic signal phase of the opposite entrance lane into the same signal phase stage, wherein the green time of the phase stage is min (disT _ LT, disT _ LT '); a superposition phase stage is added in the opposite-direction entrance lane, the opposite-direction entrance is allowed to simultaneously release in the straight-going passing direction and the left-turning passing direction, and the green light time in the superposition phase stage is | disT _ LT-disT _ LT' |;

(4) if | disT _ ST-disT _ ST '| > xi and | disT _ LT-disT _ LT' | is less than or equal to xi, the FPGA main control module coordinates with the opposite FPGA main control module through the communication module:

combining the left-turn traffic signal phase of the entrance lane and the left-turn traffic signal phase of the opposite entrance lane into the same signal phase stage, wherein the green time of the phase stage is max (disT _ LT, disT _ LT');

if disT _ ST > disT _ ST ', and maxVOL _ LT (T + T _ GAP) > maxVOL _ ST ' (T + T _ GAP), merging the straight traffic signal phase of the entrance lane and the straight traffic signal phase of the opposite entrance lane into the same signal phase stage, wherein the green time of the phase stage is min (disT _ ST, disT _ ST '); a phase superposition stage is added in an entrance lane, the straight traffic direction and the left-turn traffic direction in the entrance direction are allowed to be released simultaneously, and the green light time in the phase superposition stage is | disT _ ST-disT _ ST' |;

if disT _ ST > disT _ ST ', and maxVOL _ LT (T + T _ GAP) < maxVOL _ ST' (T + T _ GAP), merging the straight traffic signal phase of the entrance lane and the straight traffic signal phase of the opposite entrance lane into the same signal phase stage, wherein the green time of the phase stage is disT _ ST;

if disT _ ST < disT _ ST ', and maxVOL _ ST (T + T _ GAP) > maxVOL _ LT ' (T + T _ GAP), merging the left-turning traffic signal phase of the entrance lane and the left-turning traffic signal phase of the opposite entrance lane into the same signal phase stage, wherein the green time of the phase stage is disT _ LT ';

if disT _ ST < disT _ ST ' and maxVOL _ ST (T + T _ GAP) < maxVOL _ LT ' (T + T _ GAP), merging the straight traffic signal phase of the entrance lane and the straight traffic signal phase of the opposite entrance lane into the same signal phase stage, wherein the green time of the phase stage is min (disT _ ST, disT _ ST '); adding a phase superposition stage to the opposite inlet lane, allowing the straight traffic direction and the left-turn traffic direction of the inlet direction to simultaneously pass, wherein the green light time of the phase superposition stage is | disT _ ST-disT _ ST' |;

wherein: a ξ phase stage splitting critical threshold; maxVOL _ ST (T + T _ GAP), maxVOL _ ST' (T + T _ GAP) are the next prediction interval T _ GAP at time T, respectively, the straight traffic direction of the entrance lane and the maximum straight traffic lane flow rate to the straight traffic direction of the entrance lane; maxVOL _ LT (T + T _ GAP), maxVOL _ LT' (T + T _ GAP) are the next prediction interval T _ GAP at time T, left turn traffic direction of the entrance lane and the maximum left turn traffic flow to the left turn traffic direction of the entrance lane, respectively;

step5: the FPGA main control module and other FPGA main control modules at the intersection coordinate and determine the execution sequence of each signal phase, namely the phase sequence, according to the green light duration of each passing phase in each import direction calculated by each FPGA main control module, a Webster intersection delay calculation formula and the principle of minimum intersection delay by adopting the principle of minimum intersection delay through the communication module;

step6: the DSP safety monitoring module monitors the running conditions of vehicles and pedestrians in real time:

step6.1: during the period that the green light of vehicle signal is about to end, DSP safety monitoring module real-time supervision import direction is about to drive into the vehicle at intersection, calculates whether current vehicle can safely park or safely pass through the intersection:

wherein: x ₀ The minimum safe parking distance for the vehicle to be decelerated and parked; x _C The maximum safe distance for the vehicle to pass through in an accelerating way; v ₀ The running speed of a vehicle about to enter the intersection; σ is the driver's average brake reflection time; alpha is alpha _{_} Is the vehicle average braking deceleration; alpha is alpha ₊ For average braking of vehiclesSpeed; greenT is the remaining transit time; w is the width of the intersection; l is the average vehicle length;

step6.2: if X ₀ ＞X _C If the vehicle can not safely pass or safely stop, the DSP safety monitoring module is coordinated with the FPGA main control module to prolong the green time of the vehicle in the running direction so that the vehicle can safely pass, and a DilemmaType1 instruction data packet is sent to the traffic signal driving control module;

step6.3: if X ₀ ＜X _C If the traffic signal driving control unit is in a traffic signal driving state, the DSP safety monitoring module coordinates with the traffic signal driving control unit to flicker a signal lamp in the vehicle running direction, prompts a driver to decelerate and stop the vehicle or accelerate to pass through, and sends a DilemmaType2 instruction data packet to the traffic signal driving control module;

step6.4: the DSP safety monitoring module monitors the running tracks of pedestrians and vehicles in the direction of the entrance in real time, and once the situation that the running tracks of the pedestrians and the vehicles conflict with each other is detected, the DSP safety monitoring module sends danger to the traffic signal driving control unit, please note safety! "the security alert DangerInfo command packet;

step6.5: when the vehicle signal green light is about to end, the DSP safety monitoring module monitors whether pedestrians exist in the pedestrian crossing in real time, and once the pedestrians are detected to appear at the boundary of the pedestrian crossing at the edge of the road, the DSP safety monitoring module sends' patience waiting, please return to the waiting area! "safety precaution PedWarningType1 instruction data packet;

step6.6: during the period of red light of vehicle signal and green light of pedestrian signal, the DSP safety detection module monitors whether the pedestrian walks on the pedestrian crossing in real time, and once the fact that the walking track of the pedestrian is not on the pedestrian crossing is detected, the DSP safety detection module sends' civilized traffic, please walk the zebra crossing! "PedWarningType 2 instruction data packet of safety precaution;

step7: the FPGA main control module and the DSP safety detection module transmit a traffic signal control scheme including green light time of each signal phase, an execution sequence of the signal phases and a traffic safety early warning instruction to a traffic signal driving control unit through an EMIF bus.

3. The 3D lidar based self-sensing interactive traffic signal control apparatus of claim 1, wherein:

the traffic signal driving control unit consists of a traffic signal display module, a traffic signal prompt module, a traffic safety early warning module, a signal lamp set, an LED micro display screen and a 6UVPX switch signal interface board, wherein the traffic signal display module, the traffic signal prompt module, the traffic safety early warning module, the signal lamp set and the LED micro display screen are connected with the 6U VPX switch signal interface board through serial ports and are in data communication transmission; the traffic signal display module is responsible for driving and controlling the turn-on time, the turn-on state and the turn-on patterns of the signal lamp group according to the traffic signal control scheme transmitted by the traffic signal coordination unit; the traffic signal prompt module controls the information prompt content of the LED micro display screen according to the traffic signal control scheme transmitted by the traffic signal interaction coordination unit, and carries out information prompt and dynamic guidance on the traffic state and the traffic time of people and traffic; the traffic safety early warning module carries out safety early warning on pedestrian flow and traffic flow at the intersection according to the traffic safety early warning instruction transmitted by the traffic signal interaction and coordination unit;

step1: the traffic signal display module receives the traffic signal control scheme transmitted by the traffic signal interaction coordination unit, converts the digital information of the traffic signal control scheme into analog signals, and drives and controls the lighting time, lighting color, lighting sequence, transition sequence and transition form of different lamp colors and the graph displayed by the lamp colors;

step2: the traffic signal prompt module receives a traffic signal control scheme transmitted by the traffic signal interaction coordination unit, drives and controls the LED micro-display screen, and prompts and guides the traffic flow, the traffic state, the traffic time, the remaining time and the advancing direction;

step3: the traffic safety early warning module receives a traffic safety early warning instruction transmitted by the traffic signal interaction coordination unit, and carries out safety early warning according to different signal instructions:

step3.1: when the traffic safety early warning module receives a DilemmaType1 instruction data packet, taking over the control of the traffic signal display module on the signal lamp group, prolonging the green lamp duration in the appointed passing direction according to the DilemmaType1 instruction content, releasing the control on the signal lamp group when the prolonging time is over, and returning the control right of the signal lamp group to the traffic signal display module;

step3.2: when the traffic safety early warning module receives a DilemmaType2 instruction data packet, the traffic safety early warning module takes over the control of the traffic signal display module on the signal lamp group, carries out flicker control on the signal lamp in the appointed passing direction according to the DilemmaType2 instruction content, releases the control on the signal lamp group after the flicker control is finished, and returns the control right of the signal lamp group to the traffic signal display module;

step3.3: when the traffic safety early warning module receives the DangerInfo instruction data packet, the traffic signal display module takes over the control of the signal lamp group, the red light flashing early warning is started in the appointed traffic direction according to the DangerInfo instruction content, and the danger is alarmed through the voice and the LED micro display screen, please pay attention to the safety! ";

step3.4: when the traffic safety early warning module receives a PedWarningType1 instruction data packet, the pedestrian is warned of' waiting for patience, please return to the waiting area! ";

step3.5: when the traffic safety early warning module receives a PedWarningType2 instruction data packet, the pedestrian is warned of' civilized traffic, please walk the zebra crossing!through the voice and the LED micro display screen! ".

Images