CN107798861A - A kind of vehicle cooperative formula formation running method and system - Google Patents

Abstract

The invention belongs to the field of vehicle network-connected automatic driving, and discloses a vehicle cooperative formation driving method and system. By capturing vehicle perception information and vehicle absolute position information, perception fusion, path planning, and decision-making control are performed; and vehicle roads are integrated. Environmental information, realize the vehicle formation vehicle convergence, car following, exit, parking and cooperative lane change cooperative driving strategy. The invention uses 5G to realize the connection of each module and direct communication between modules, and at the same time installs low-cost on-board sensors such as vision and millimeter-wave radar, integrates perception fusion, path planning, and decision-making control into the same software architecture, and integrates the vehicle road environment Information, through the high-precision controller to achieve reliable vehicle formation driving function. The system conforms to the development trend of networked vehicle platooning. It is highly modular and easy to expand functions. The decentralized car-following strategy can reduce the impact of network communication uncertainty.

Description

Method and system for vehicle cooperative formation driving

technical field

The invention belongs to the field of vehicle network-connected automatic driving, and in particular relates to a vehicle cooperative formation driving method and system.

Background technique

Intelligent networked vehicles refer to equipped with advanced on-board sensors, controllers, actuators and other devices, and integrate modern communication and network technologies to realize the exchange and sharing of intelligent information between vehicles and X (people, vehicles, roads, clouds, etc.), with complex Functions such as environmental perception, intelligent decision-making, and collaborative control can realize "safe, efficient, comfortable, and energy-saving" driving, and finally realize a new generation of vehicles that can replace humans to operate. Vehicle intelligence can be divided into five levels: driver assistance (DA), partially automated driving (PA), conditional automated driving (CA), highly automated driving (HA), and fully automated driving (FA).

Networking can be divided into three levels: network-assisted information interaction, network collaborative perception, and network collaborative decision-making and control. With the maturity of vehicle communication and computing technology, the research on fleet control technology of intelligent vehicle road system has become a hot spot. The intelligent vehicle road system forms one or more fleets with common speed and small distance between vehicles, which effectively reduces traffic accidents caused by human factors and enhances traffic safety; at the same time, it can reduce vehicle exhaust emissions and environmental pollution.

The current international mainstream fleet cooperative driving system structure includes network layer, link layer, coordination layer, control layer and physical layer. The cooperative driving strategy mainly includes formation vehicle convergence, car following, exit, parking and cooperative lane change. Due to the uncertainty of the communication environment, the car-following strategy is selected according to the status information it sends, and the robustness of the formation model is not strong.

In summary, the problems in the prior art are:

The current network-connected vehicle formation algorithm can only be used on structured roads with high-precision maps, and the leading vehicle (lead vehicle) is the center, and the car-following strategy is selected according to the status information sent by it. The robustness of the formation model is not strong ; Not combined with the method based on control stability theory to provide a kind of decentralized self-driving multi-vehicle cooperative formation driving.

Contents of the invention

Aiming at the problems existing in the prior art, the present invention provides a vehicle cooperative formation driving method and system.

The present invention is achieved in this way, a vehicle cooperative formation driving method, the vehicle cooperative formation driving method performs perception fusion, path planning, decision-making control by capturing vehicle perception information and vehicle absolute position information; and integrates vehicle roads Environmental information, realize the vehicle formation vehicle convergence, car following, exit, parking and cooperative lane change cooperative driving strategy.

Further, the vehicle cooperative formation driving method specifically includes:

Step 1. Use a dedicated experimental network to build a network layer architecture, and use 5G to realize direct communication between the central service unit CSU, roadside unit RSU, and vehicle-mounted unit OBU at the network layer;

Step 2, using on-board sensors and high-precision maps to collect vehicle positioning module information;

Step 3: Use the human-computer interaction module HMI of the coordination layer to select the end point of formation driving, use the central service unit CSU to issue the global planning path, and the roadside unit RSU transparently transmits it to the vehicle-mounted unit OBU to solve the local planning path;

Step 4, use the human-computer interaction module HMI of the coordination layer to create a formation;

Step 5: After the formation is created, the formation command is issued to the control layer; the control field is filled in according to the formation status of vehicle convergence, vehicle following, formation exit, formation parking, and cooperative lane change to realize real-time control;

Step 6: The controller of the control layer performs speed control according to the status information of the vehicle in front obtained by the on-board sensor, and completes the vehicle convergence;

Step 7, when the vehicles converge and reach the target distance, enter the vehicle following state;

Step 8, when the vehicle exits the formation halfway, enter the formation vehicle exit state;

Step 9: Before the formation reaches the key points of the overall planning, enter the parking state of the formation vehicles.

Further, in step 1, the on-board unit OBU communicates with a wireless router through a wireless network; the wireless router communicates with a central service unit CSU and a roadside unit RSU through an Ethernet network;

Step 2 specifically includes: installing an on-board sensor including a perception module; the perception module includes vision and millimeter wave radar sensors; installing a positioning module including GPS\inertial navigation and high-precision maps, and the GPS\inertial navigation is used to receive The GPS signal and the vehicle sensor signal are used to calculate the absolute position information of the own vehicle; the vehicle controller is installed, and the vehicle control obtains the communication data of the perception module and the positioning module through the CAN bus and the Ethernet respectively, and interacts with the network layer, At the same time, the traffic information provided by the roadside system is obtained to form a link layer.

Further, step three specifically includes:

At the coordination layer, the destination of the guided vehicle formation is selected through the human-computer interaction interface HMI, and after reporting to the network layer, the central service unit CSU makes global path planning, and sends high-precision map scripts to the guided vehicles to define the path, and the on-board unit OBU obtains After the path planning key points are listed, the current starting point is used for re-planning, and a smooth centerline of the road is generated every 1m, and finally a local path plan including key points is obtained, which is called by the vehicle control layer.

Further, Step 4 specifically includes: guiding the vehicle to select the vehicle ID that joins the formation in the HMI selection of the human-machine interface, and sending a formation creation request through network communication, and after the vehicles in the same lane accept the request, calculate their own position through the on-board unit OBU Information, vehicle driving environment information obtained by on-board sensors, detection and identification of traffic signs, road markings and dynamic and static obstacles, after multi-source information fusion processing, request or refuse to join the formation; guide vehicles to confirm the formation vehicles within a fixed period information on the number and location, and send path planning information to the formation vehicles to create formations.

Further, in step 6, when the vehicles in the formation converge, the following control input model is used

Obtain the relative position x _i , velocity v _i , and acceleration a _i of the vehicle through GPS\inertial navigation, and calculate the relative position x _i-1 , velocity v _i-1 , and Acceleration a _i-1 front vehicle control target acceleration Obtained by the network layer, and finally solve the control target of the vehicle The output is done through the control layer of the vehicle;

In step 7, when the formation vehicles are following, set the target following distance Δd _tar as a certain value, set Δd=x _{i-1 -xi} , which is the current actual distance of formation vehicles, when |Δd _tar -Δd|≤1m v _i ＝v _i-1 , the vehicle keeps following the car, otherwise go to step 6;

In step 8, when the vehicle exits the formation midway, it sends a request to the leading vehicle. When the exit is permitted, the vehicle detects whether there are dynamic and static obstacles and traffic signs in the adjacent lane through the camera and the vehicle-mounted side millimeter-wave radar, and judges the current situation through information fusion decision-making. Whether the road environment can change lanes; after the on-board unit OBU calculates the lane change, the vehicle switches to the single-vehicle automatic driving mode, and fills the lane-changing instruction field to the on-board controller. Update the number and status information of the vehicles in the formation, and return to step six.

Further, in step 9, judge the distance between the end point of the global path planning script and the current approach point of the guided vehicle. When the distance is less than 100m, enter the formation vehicle parking mode; guide the vehicle to set the following time distance according to its own speed, so that Δd _tar = th*v _l +Δd _safe , after the time distance parameter th is set, the distance of the formation vehicles will decrease as the speed decreases, and the safe distance to be kept at the final stop is Δd _safe .

Further, the vehicle cooperative formation driving method also includes:

Fleet cooperative lane change, when vehicles on adjacent roads find a driving convoy, they send a joining request to the leading vehicle, and the guiding vehicle accepts/refuses the joining request according to the maximum number of vehicles in the convoy and the formation status limit; when the vehicle is allowed to join the convoy, Calculate the joining point, and send the IDs of the vehicles before and after the joining point to the leading vehicle at the same time, set the longitudinal speed to the controller of the joining vehicle with the vehicle in front of the joining point as the car-following target, and gradually approach the joining point; The vehicle sends a lane-changing request, and the joining vehicle invokes the fleet convergence algorithm at this time, and the vehicle behind the joining point sets the target car-following distance to 2Δd _tar to increase the distance between the lane-changing vehicles and the preceding vehicle; when the target distance is reached, The leading vehicle sends a lane-changing command to the joining vehicle. After the lane change is successful, the vehicle behind the joining point changes the following target to the joining vehicle, returns to the convoy state, and the leading vehicle updates the number of vehicles in the convoy and status information.

Another object of the present invention is to provide a vehicle cooperative formation driving system.

Advantage of the present invention and positive effect are:

The invention is developed for conditional automatic driving (CA) working conditions, and conforms to the current trend of network-connected vehicles driving in formation; the system structure is clear, highly modularized, and convenient for fleet function expansion; decentralized, formation vehicles do not have to follow the car Keep abreast of the status of the leading vehicle (lead vehicle) at all times.

The invention uses 5G to realize the connection of each module and direct communication between modules, and at the same time installs low-cost on-board sensors such as vision and millimeter-wave radar, integrates perception fusion, path planning, and decision-making control into the same software architecture, and integrates the vehicle road environment Information, through the high-precision controller to achieve reliable vehicle formation driving function. The system conforms to the development trend of networked vehicle platooning. It is highly modular and easy to expand functions. The decentralized car-following strategy can reduce the impact of network communication uncertainty.

Description of drawings

Fig. 1 is a flowchart of a vehicle cooperative formation driving method provided by an embodiment of the present invention.

Fig. 2 is a flow chart of team cooperative lane changing provided by an embodiment of the present invention.

Fig. 3 is a network design block diagram of the vehicle cooperative formation driving system provided by the embodiment of the present invention.

Fig. 4 is a block diagram of a vehicle cooperative formation driving system provided by an embodiment of the present invention.

Detailed ways

In order to make the object, technical solution and advantages of the present invention more clear, the present invention will be further described in detail below in conjunction with the examples. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention.

Software of the present invention is based on linux (Ubuntu) operating system development, collocation high-performance computer, center service unit (CSU) software interior uses LCM interface (public open source), external communication uses ZeroMQ (public open source), development language C, C++.

The application principle of the present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

As shown in Figure 1, the vehicle cooperative formation driving method provided by the embodiment of the present invention includes:

Step 1: Use a dedicated experimental network to build a network layer architecture, and use 5G to realize the connection of each module. These modules mainly include a central service unit (CSU), a roadside unit (RSU), and an on-board unit (OBU). The modules communicate directly .

Step 2: Add on-board sensors. The perception module mainly includes low-cost sensors such as vision and millimeter-wave radar (monocular cameras and stereo cameras are used for vision sensors, and millimeter-wave radars include forward long-range millimeter-wave radars and four-corner mid-range millimeter-wave radars. ), the positioning module, including GPS\inertial navigation and high-precision map information, GPS\inertial navigation receives GPS signals and vehicle sensor signals, and calculates the absolute position information of the own vehicle. The on-board controller obtains the communication data of the perception module and the positioning module through the CAN bus and Ethernet respectively, and interacts with the network layer, and at the same time obtains the traffic information provided by the roadside system to form a link layer.

Step 3: At the coordination layer, select the driving destination of the leading vehicle (lead vehicle) in formation through the Human-Machine Interface (HMI), and report to the network layer, the central service unit (CSU) makes global path planning, and issues The high-precision map script defines the path (including the starting point, key passing point, end point, and latitude and longitude information). After the on-board unit (OBU) obtains the list of key points for path planning, it uses the current starting point for re-planning, and generates a road smoothing center approximately every 1m Lines, and finally a local path plan including key points is obtained, which is invoked by the vehicle's control unit.

Step 4: The coordination layer starts to create a formation, guides the vehicle to select the vehicle ID (license plate number) that needs to join the formation on the human-machine interface (HMI), and sends a formation creation request through network communication. After the vehicles in the same lane accept the request, they pass The on-board unit (OBU) calculates its own position information, detects and recognizes traffic signs, road markings, and dynamic and static obstacles through the vehicle driving environment information obtained by sensors such as monocular cameras, stereo cameras, and millimeter-wave radars, and performs multi-source information fusion After processing, request or decline to join the formation. The guiding vehicle confirms the number and location of vehicles joining the formation within a fixed period, and sends route planning information to the formation vehicles, and the formation is created.

Step 5: After the coordination layer creates the formation, it sends formation instructions to the control layer. For real-time control, the control field must be populated according to the formation state. The formation state defined by the present invention includes vehicle convergence, vehicle following, exit formation, formation parking, and coordinated lane change.

Step 6: When the formation vehicles converge, use the following control input model

Obtain the relative position x _i , velocity v _i , and acceleration a _i of the vehicle through GPS\inertial navigation, and calculate the relative position x _i-1 , velocity v _i-1 , and Acceleration a _i-1 front vehicle control target acceleration Obtained by the network layer, and finally solve the control target of the vehicle The output is done through the underlying control actuators of the vehicle.

Step 7: When the formation vehicles are following, set the target following distance Δd _tar (a certain value), set Δd=x _i-1 -x _i , which is the current actual formation vehicle distance, when |Δd _tar -Δd|≤1m When v _i =v _i-1 , the vehicle keeps following the car, otherwise go to step 6;

Step 8: Formation vehicle exits. When the vehicle exits the formation halfway, it sends a request to the leading vehicle (lead vehicle). When the vehicle is allowed to exit, the vehicle detects whether there are dynamic and static obstacles in the adjacent lane through the camera and the vehicle-mounted side millimeter-wave radar. Traffic signs , through information fusion decision-making to determine whether the current road environment can change lanes; after the on-board unit (OBU) can change lanes after calculation, the vehicle switches to single-vehicle automatic driving mode, and fills the lane change instruction field to the controller, and the vehicle lane change is completed and sent to the The leading vehicle (leading vehicle) sends an exit success message, the leading vehicle (leading vehicle) updates the number and status information of the formation vehicles, and returns to step six.

Step 9: Park the vehicles in the formation, judge the distance between the end point of the global path planning script and the current approach point of the leading vehicle (lead vehicle), and enter the parking mode of the formation vehicles when the distance is less than 100m. Guide the vehicle to set the following time distance according to its own speed, that is, Δd _tar =th*v _l +Δd _safe , after the parameter th (time distance) is set, the distance of the formation vehicles will decrease as the speed decreases, and the final stop will be kept at a safe The distance Δd _safe .

As shown in Figure 2, the team cooperative lane change provided by the embodiment of the present invention, when the adjacent road vehicle finds a driving team, it sends a joining request to the leading vehicle (lead vehicle), and the guiding vehicle is based on the maximum number of vehicles in the team and formation Status limits accepting\rejecting join requests. When the vehicle is allowed to join the fleet, the joining point is calculated, and the ID (license plate number) of the vehicle before and after the joining point is sent to the head car at the same time. Join points close. When arriving at the joining point, send a lane-changing request to the leading vehicle, and the joining vehicle will call the fleet convergence algorithm at this time, and the vehicle behind the joining point will set the target car-following distance to 2Δd _tar at this time, so as to separate the distance between the two lane-changing vehicles and the vehicle in front. Distance; when the target distance is reached, the guiding vehicle sends a lane change instruction to the joining vehicle. After the lane change is successful, the vehicle behind the joining point changes the following target to the joining vehicle, returns to the convoy state, and the leading vehicle (leading vehicle) updates the fleet Number of vehicles, status information.

The application principle of the present invention will be further described below in combination with specific system and network design.

As shown in FIG. 4 , the vehicle cooperative formation driving system provided by the embodiment of the present invention includes a network layer, a link layer, a coordination layer, a control layer and a physical layer.

The main module of the network layer. Including central service unit (CSU), roadside unit (RSU), on-board unit (OBU). Use VPN to realize direct communication between modules. To facilitate network deployment, put CSU and RSU in a local area network and connect them with a wireless router (refer to Figure 1). Since the OBU, RSU and CSU all adopt a software design framework of several independent processes, and the inter-process communication adopts the LCM mode, in order to keep the overall structure unchanged, the overall communication is divided into two categories: one is the communication within the network element, that is, the communication within the local area network , which is implemented by each entity unit using LCM; the other type is communication between network elements, that is, communication between different network entities. Since LCM can only work in a local area network and cannot work across network segments, the ZeroMQ communication method is used. Realization, network design block diagram shown in Figure 3.

The vehicle positioning module includes GPS\inertial navigation and high-precision map information, GPS\inertial navigation receives GPS signals and vehicle sensor signals, solves the absolute position information of the own vehicle, and sends the position information to the automatic driving through Ethernet Controller and on-board unit (OBU); high-precision map stores relevant information such as road edges, lane centerlines, tunnels, road section ramps, road section curvature, ramp entrances and exits, service area entrances and exits, traffic sign locations and landmark buildings etc. Absolute position information.

The human-computer interaction interface (HMI) of the formation receives the starting point and the end point set by the driver in the human-computer interaction system, and completes the global path planning of the automatic driving behavior; and combines the driving route planned by the global path planning and the positioning of the vehicle fusion positioning module Information, on the vehicle driving local environment map established by the multi-sensor information fusion module, plan a local driving route that meets the traffic rules and safety requirements.

The vehicle convergence control model of the formation, the vehicle state model is described as follows:

τ _i is the engine time constant, which is a parameter related to the car, is the last control output. The amount of error controlled in this way can be described as:

δ _i ＝x _i-1 -x _i -H _i

Where H _i =Δd _tar is the target inter-vehicle distance. The present invention introduces an expected space error and is specifically described as:

Here t _go represents the expected distance achieved by keeping the t _go time from the current moment; the present invention can construct a lyapunov function whose original function is positive and whose derivative is strictly decreasing by applying the lyapunov direct method, specifically defined as:

After the simultaneous equations, the control output model can be obtained:

The on-board computing unit will control the output signal Feedback to the vehicle execution unit, and finally the vehicle speed control is completed under the joint action of the throttle opening and the braking force, and the formation vehicle convergence function is realized.

When the formation vehicles are following, set the target following distance Δd _tar , and the on-board computing unit calculates Δd= _xi-1 -xi _i according to the current vehicle state information, which is the current actual distance of formation vehicles; when |Δd _tar -Δd|≤ At 1m, v _i =v _i-1 , transparently transmit the speed control target to the bottom execution unit as the speed of the vehicle in front, and the vehicle enters the following state, otherwise enters the vehicle convergence state.

When the formation vehicle exits, the exit vehicle sends a request to the leading vehicle (lead vehicle) (the leading vehicle is not allowed to exit the formation, and the single-lane exit is not allowed). When the exit is permitted, the vehicle detects whether there are dynamic and static obstacles and traffic signs in the adjacent lane through the camera and the vehicle-mounted lateral milliwave radar, and judges whether the current road environment can change lanes through information fusion decision-making; the on-board unit (OBU) can be calculated After changing lanes, the vehicle switches to single-vehicle automatic driving mode, and fills the lane-changing instruction field to the controller. After the vehicle lane change is completed, the exit success message is sent to the leading vehicle (leading vehicle), and the route is re-planned. The leading vehicle (leading vehicle) updates the number and status information of the formation vehicles, and the fleet enters the converging state, re-locking the leading vehicle.

When the formation vehicles stop, judge the distance between the end point of the global path planning script and the current nearest driving path point of the leading vehicle (lead vehicle). When the distance is less than 100m, enter the formation vehicle parking mode. Guide the vehicle to set the following time distance according to its own speed, that is, Δd _tar = th*v _l + Δd _safe , after the parameter th is set, the distance of formation vehicles will decrease with the decrease of speed, and finally keep at a safe distance Δd _safe when stopping. When Δd= _{xi-1- xi} is smaller than Δd _safe , the vehicle is braked.

The flow chart of the fleet control methods such as gathering, following, exiting, and parking of formation vehicles is shown in Figure 1.

The roadside unit (RSU) sends formation information (position, speed, angle, acceleration, locking relationship) to the on-board unit (OBU) at 1Hz, and the on-board unit (OBU) transparently transmits to the human-machine interface (HMI), and the HMI displays the formation Information, events are reported in real time (including braking).

Coordinated lane change of platoon vehicles: When vehicles on adjacent roads find a convoy that is driving, they send a request beacon to the leading vehicle (lead vehicle) to join the convoy, and start timer 1 at the same time. Status (whether the system is busy or not), limit acceptance/rejection of joining requests, and continuously update the joining vehicle information (vehicle ID, fleet number) before timer 1 expires. After the vehicle is allowed to join the convoy, cancel timer 1, guide the vehicle (lead vehicle) to set the joining point, and set the longitudinal speed and joining time to the longitudinal controller of the target vehicle. Judging whether the current point of the joining vehicle is the target following vehicle, if not following the vehicle, then the leading vehicle (lead vehicle) sends a request for expanding the distance GAP_DIS to the non-following vehicle, where the target following distance DIS_REQ is set to 2Δd _tar ; if the vehicle ahead is joining The target of the vehicle follows the vehicle, and when GAP_DIS≥DIS_REQ, update the fleet status information, start the vehicle aggregation algorithm, request to join the vehicle to complete the coordinated lane change, and the formation will update the ID and formation number after the lane change again, and the information flag will be set to ACK (indicating all information already received). When the lane change instruction is sent and the system is busy, set the beacon Busy=true, and start timer 3 at the same time, after receiving the coordinated lane change completion beacon, cancel timer 3, update the number of vehicle formations, vehicle list, and signaling If the standard Busy=false, the vehicle cooperative lane changing action in this period is completed.

The flowchart of the cooperative lane-changing control method for formation vehicles is shown in Fig. 2 .

The software of the present invention is developed based on the linux (Ubuntu) operating system, the development language C, C ++, and a high-performance computer is matched. The deployment position of the central service unit (CSU) can select a suitable position according to the principle of network load balancing, which is usually a regional center. The CSUs in each regional center are connected by wire to form a connectivity graph to achieve global traffic control. The central service unit (CSU) is connected to the roadside unit (RSU) through Ethernet.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present invention should be included in the protection of the present invention. within range.

Claims (9)

1. A vehicle cooperative formation driving method, characterized in that, the vehicle cooperative formation driving method performs perception fusion, path planning, decision-making control by capturing vehicle perception information and vehicle absolute position information; and integrates the vehicle road environment Information, to realize the vehicle formation vehicle convergence, car following, exit, parking and cooperative lane change cooperative driving strategy. 2. The vehicle cooperative formation traveling method according to claim 1, wherein the vehicle cooperative formation traveling method specifically comprises: Step 1. Use a dedicated experimental network to build a network layer architecture, and use 5G to realize direct communication between the central service unit CSU, roadside unit RSU, and vehicle-mounted unit OBU at the network layer; Step 2, using on-board sensors and high-precision maps to collect vehicle positioning module information; Step 3: Use the human-computer interaction module HMI of the coordination layer to select the end point of formation driving, use the central service unit CSU to issue the global planning path, and the roadside unit RSU transparently transmits it to the vehicle-mounted unit OBU to solve the local planning path; Step 4, use the human-computer interaction module HMI of the coordination layer to create a formation; Step 5: After the formation is created, the formation command is issued to the control layer; the control field is filled in according to the formation status of vehicle convergence, vehicle following, formation exit, formation parking, and cooperative lane change to realize real-time control; Step 6: The controller of the control layer performs speed control according to the status information of the vehicle in front obtained by the on-board sensor, and completes the vehicle aggregation; Step 7, when the vehicles converge and reach the target distance, enter the vehicle following state; Step 8, when the vehicle exits the formation halfway, enter the formation vehicle exit state; Step 9: Before the formation reaches the key points of the overall planning, enter the parking state of the formation vehicles. 3. The vehicle cooperative formation driving method according to claim 2, wherein in step 1, the on-board unit OBU communicates with the wireless router through the wireless network; the wireless router communicates with the central service unit CSU through the Ethernet respectively , Road side unit RSU communication; Step 2 specifically includes: installing an on-board sensor including a perception module; the perception module includes vision and millimeter wave radar sensors; installing a positioning module including GPS\inertial navigation and high-precision maps, and the GPS\inertial navigation is used to receive The GPS signal and the vehicle sensor signal are used to calculate the absolute position information of the own vehicle; the vehicle controller is installed, and the vehicle control obtains the communication data of the perception module and the positioning module through the CAN bus and the Ethernet respectively, and interacts with the network layer, At the same time, the traffic information provided by the roadside system is obtained to form a link layer. 4. The vehicle cooperative formation driving method according to claim 2, wherein step 3 specifically comprises: At the coordination layer, the destination of the guided vehicle formation is selected through the human-computer interaction interface HMI, and after reporting to the network layer, the central service unit CSU makes global path planning, and sends high-precision map scripts to the guided vehicles to define the path, and the on-board unit OBU obtains After the path planning key points are listed, the current starting point is used for re-planning, and a smooth centerline of the road is generated every 1m, and finally a local path plan including key points is obtained, which is called by the vehicle control layer. 5. The vehicle cooperative formation driving method according to claim 2, wherein step 4 specifically comprises: guiding the vehicle to join the vehicle ID of the formation in the HMI selection of the human-machine interface, and sending a formation formation request through network communication, After receiving the request, the vehicles in the same lane will calculate their own position information through the on-board unit OBU, and use the vehicle driving environment information obtained by the on-board sensors to detect and identify traffic signs, road markings and dynamic and static obstacles, and perform multi-source information fusion processing After that, request or refuse to join the formation; guide the vehicle to confirm the number and location information of the vehicles joining the formation within a fixed period, and send path planning information to the formation vehicles to create the formation. 6. The vehicle cooperative formation driving method according to claim 2, wherein in step 6, when the formation vehicles gather, the following control input model is adopted <mrow><msubsup><mi>a</mi><mi>i</mi><mi>c</mi></msubsup><mo>=</mo><mi>f</mi><mrow><mo>(</mo><msub><mi>x</mi><mrow><mi>i</mi><mo>-</mo><mn>1</mn></mrow></msub><mo>,</mo><msub><mi>v</mi><mrow><mi>i</mi><mo>-</mo><mn>1</mn></mrow></msub><mo>,</mo><msub><mi>a</mi><mrow><mi>i</mi><mo>-</mo><mn>1</mn></mrow></msub><mo>,</mo><msub><mi>x</mi><mi>i</mi></msub><mo>,</mo><msub><mi>v</mi><mi>i</mi></msub><mo>,</mo><msub><mi>a</mi><mi>i</mi></msub><mo>,</mo><msubsup><mi>a</mi><mrow><mi>i</mi><mo>-</mo><mn>1</mn></mrow><mi>c</mi></msubsup><mo>)</mo></mrow><mo>;</mo></mrow> Obtain the relative position x _i , velocity v _i , and acceleration a _i of the vehicle through GPS\inertial navigation, and calculate the relative position x _i-1 , velocity v _i-1 , and Acceleration a _i-1 front vehicle control target acceleration Obtained by the network layer, and finally solve the control target of the vehicle The output is done through the control layer of the vehicle; In step 7, when the formation vehicles are following, set the target following distance Δd _tar as a certain value, set Δd=x _{i-1 -xi} , which is the current actual distance of formation vehicles, when |Δd _tar -Δd|≤1m v _i ＝v _i-1 , the vehicle keeps following the car, otherwise go to step 6; In step 8, when the vehicle exits the formation midway, it sends a request to the leading vehicle. When the exit is permitted, the vehicle detects whether there are dynamic and static obstacles and traffic signs in the adjacent lane through the camera and the vehicle-mounted side millimeter-wave radar, and judges the current situation through information fusion decision-making. Whether the road environment can change lanes; after the on-board unit OBU calculates the lane change, the vehicle switches to the single-vehicle automatic driving mode, and fills the lane-changing instruction field to the on-board controller. Update the number and status information of the vehicles in the formation, and return to step six. 7. The vehicle cooperative formation driving method according to claim 2, wherein in step 9, the distance between the end point of the global path planning script and the current approach point of the guiding vehicle is judged, and when the distance is less than 100m, the vehicle enters the formation Parking mode: guide the vehicle to set the following time distance according to its own speed, so that Δd _tar =th*v _l +Δd _safe , after the time distance parameter th is set, the distance of the formation vehicles will decrease as the speed decreases, and the final distance maintained when parking The safe distance is Δd _safe . 8. The vehicle cooperative formation driving method according to claim 2, wherein the vehicle cooperative formation driving method further comprises: Fleet cooperative lane change, when vehicles on adjacent roads find a driving convoy, they send a joining request to the leading vehicle, and the guiding vehicle accepts/refuses the joining request according to the maximum number of vehicles in the convoy and the formation status limit; when the vehicle is allowed to join the convoy, Calculate the joining point, and send the IDs of the vehicles before and after the joining point to the leading vehicle at the same time, set the longitudinal speed to the controller of the joining vehicle with the vehicle in front of the joining point as the car-following target, and gradually approach the joining point; The vehicle sends a lane-changing request, and the joining vehicle invokes the fleet convergence algorithm at this time, and the vehicle behind the joining point sets the target car-following distance to 2Δd _tar to increase the distance between the lane-changing vehicles and the preceding vehicle; when the target distance is reached, The leading vehicle sends a lane-changing command to the joining vehicle. After the lane change is successful, the vehicle behind the joining point changes the following target to the joining vehicle, returns to the convoy state, and the leading vehicle updates the number of vehicles in the convoy and status information. 9. A vehicle cooperative formation driving system according to the vehicle cooperative formation driving method according to claim 1.