Lane keeping control method based on road condition

Abstract

The invention discloses a lane keeping control method based on road conditions, and the method comprises the steps: detecting the information of a lane line and the information of two sides of the lane line in real time; judging whether the lane line is kept to control the vehicle according to a lane centering control strategy or not based on the detected lane line information and the information on the two sides of the lane line; if yes, adopting a lane centering control strategy to control the vehicle, and otherwise, adopting an active deviation strategy to control the vehicle; wherein the active departure strategy means that the vehicle is controlled to run in the lane and deviate towards the two sides in the lane. The control target of the whole system is properly adjusted according to the current scene, the transverse lane keeping function cannot keep the vehicle running in the center of the lane at a glance, a reasonable path needs to be planned according to the current scene, the driving behavior of the system is more humanoid, and a user has better functional experience.

Description

Technical Field

[0001] The present invention relates to the field of intelligent assisted driving of automobiles, and in particular to an intelligent avoidance method based on road conditions. Background Art

[0002] The ADAS assisted driving system is a component of vehicle-mounted assisted driving. The lane keeping function in assisted driving is the most basic assisted driving control strategy. Its main goal is to control the vehicle to keep driving in the center of the lane based on the recognition of the lane line, thereby avoiding the safety hazards of assisted driving caused by lane line deviation. For example, the lane keeping method, lane keeping control device and lane keeping system with patent application number 202210427280.9 disclose that the lane keeping method includes: obtaining lane line information and wheel speed information; performing the lane keeping function according to the lane line information and wheel speed information. The lane keeping method provided in the embodiment of the present application can detect the lateral slope of the road on which the vehicle is traveling in real time according to the wheel speed information, and adjust the lane keeping function according to the detection results, so that the LKS function can stably control the vehicle to drive in the center of the lane and reduce the snaking caused by the influence of the road surface.

[0003] Although staying in the center of the lane line can avoid safety risks caused by deviation during vehicle driving, the existing technology only considers lane line information in the lane keeping function, and does not consider the current driving environment. For example, when driving in a lane close to the guardrail and there is a large truck next to the vehicle, the vehicle keeping the lane center may bring a greater sense of oppression to the driver, making the functional experience not so good; even when there is a large truck on the side, there will be certain scratches due to the cargo on the large truck. Therefore, the traditional lane keeping assisted driving technology solution cannot provide a safer, more reliable and better assisted driving experience. In order to give users a better functional experience, the ADAS system needs to appropriately adjust the control target of the entire system according to the current scenario. The lateral lane keeping function will not blindly keep the vehicle in the center of the lane. It is necessary to plan a reasonable path according to the current scenario to make the system driving behavior more human-like and give users a better functional experience. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a lane keeping control method based on road conditions, which appropriately adjusts the control target of the entire system according to the current scenario. The lateral lane keeping function will not blindly keep the vehicle in the center of the lane, but needs to plan a reasonable path according to the current scenario, so that the system driving behavior is more human-like and users have a better functional experience.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is: a lane keeping control method based on road conditions, comprising:

[0006] Real-time detection of lane line information and information on both sides of the lane line;

[0007] Based on the detected lane line information and the information on both sides of the lane line, it is judged whether the lane line keeping controls the vehicle according to the lane centering control strategy; if so, the lane centering control strategy is used to control the vehicle, otherwise the active deviation strategy is used to control the vehicle; the active deviation strategy refers to controlling the vehicle to drive within the lane and deviate to both sides of the lane.

[0008] Based on the information on both sides of the lane line, the obstacles and vehicle information on both sides of the lane are identified, and the vehicle is controlled to drive within the lane and deviate to the side of the lane away from obstacles or vehicles in the active deviation strategy.

[0009] In the active deviation strategy, the vehicle's driving path is planned based on the real-time detected lane line information and the information on both sides of the lane line, and the vehicle is controlled according to the driving path.

[0010] Planning the vehicle's driving path in the active departure strategy includes:

[0011] S1. Establish vehicle dynamics model;

[0012] S2. Establishing a vehicle prediction model based on the vehicle dynamics model;

[0013] S3: Divide the area in the lane based on the information on both sides of the detected lane line and establish hard constraints on the route in the lane

[0014] S4. Establish hard constraints on the lateral position in the lane and dynamic safety constraints;

[0015] S5. Solve the pre-established objective function based on the constraints in steps S3 and S4 to obtain corresponding path control parameters.

[0016] The dynamic safety constraints in S4 include constraints on the front wheel turning angle, vehicle lateral acceleration, and vehicle sideslip angle.

[0017] :The objective function established is

[0018]

[0019] The advantages of the present invention are that it can timely determine whether there is a sub-feasible area according to the current scene of the vehicle, and by solving the objective function, a control sequence is given to control the vehicle to avoid the corresponding target, thereby improving the user experience of the lane keeping function. The control target of the entire system is appropriately adjusted according to the current scene. For the lateral lane keeping function, the vehicle will not blindly keep driving in the center of the lane. It is necessary to plan a reasonable path according to the current scene, so that the system driving behavior is more human-like and the user has a better functional experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The following is a brief description of the contents expressed in the drawings of the present invention and the symbols in the drawings:

[0021] Figure 1 A schematic diagram of a vehicle driving trajectory under the lane keeping control method based on road conditions of the present invention;

[0022] Figure 2 It is a schematic diagram of the model predictive control MPC of the present invention.

[0023] 1—infeasible area;

[0024] 2—Secondary feasible area (reference points will be intercepted in it);

[0025] 3—Feasible area. DETAILED DESCRIPTION

[0026] The specific implementation of the present invention will be further explained in detail below by describing the optimal embodiment with reference to the accompanying drawings.

[0027] The main invention of this application is to improve the lane keeping control strategy of the prior art to achieve a more humanized control strategy while satisfying the lane keeping function and safety. The analysis of the current scene and the path planning that satisfies the vehicle dynamics stability can make the lane keeping function more humanized while ensuring the stability and safety of the vehicle, so that the functional experience is better. The specific solutions include:

[0028] The lane keeping control method based on road conditions includes:

[0029] Real-time detection of lane line information and information on both sides of the lane line; judging whether lane line keeping controls the vehicle according to the lane centering control strategy based on the detected lane line information and information on both sides of the lane line;

[0030] If so, the lane centering control strategy is used to control the vehicle, otherwise the active deviation strategy is used to control the vehicle; the active deviation strategy refers to controlling the vehicle to travel within the lane and deviate to both sides of the lane.

[0031] Based on the information on both sides of the lane line, the obstacles and vehicle information on both sides of the lane are identified, and the vehicle is controlled to drive within the lane and deviate to the side of the lane away from obstacles or vehicles in the active deviation strategy.

[0032] In the active deviation strategy, the vehicle's driving path is planned based on the real-time detected lane line information and the information on both sides of the lane line, and the vehicle is controlled according to the driving path.

[0033] Planning the vehicle's driving path in the active departure strategy includes:

[0034] S1. Establish vehicle dynamics model;

[0035] S2. Establishing a vehicle prediction model based on the vehicle dynamics model;

[0036] S3: Divide the area in the lane based on the information on both sides of the detected lane line and establish hard constraints on the route in the lane

[0037] S4. Establish hard constraints on the lateral position in the lane and dynamic safety constraints;

[0038] S5. Solve the pre-established objective function based on the constraints in steps S3 and S4 to obtain corresponding path control parameters.

[0039] The dynamic safety constraints in S4 include constraints on the front wheel turning angle, vehicle lateral acceleration, and vehicle sideslip angle.

[0040] The specific content of the present invention is: first, it will determine whether to execute the lane centering control strategy or the lane active deviation strategy according to the lane line information and the information on both sides of the lane line. The lane line information includes the lane line type, lane width, etc.; the lane side information includes the type of vehicles in the adjacent lane, the lateral distance, longitudinal distance, lateral relative speed, longitudinal relative speed, etc. of the adjacent lane vehicles and the vehicle. Its judgment strategy is: when it is detected that the lane width is greater than the set threshold value and one side of the vehicle is the edge of the road or it is detected that there are vehicles such as trucks or trucks in the lane on one side of the lane where the vehicle is located, and there is no vehicle in the other lane, and by judging whether the thresholds such as the relative speed and relative longitudinal distance between the vehicle and the truck meet the conditions, if so, the vehicle lane active deviation strategy is executed, and the vehicle is controlled to deviate from the lane in the lane away from the edge of the road or away from the set vehicle type such as trucks or trucks, so as to ensure safety during driving. Note that the active deviation of this application is limited to the active deviation from the middle line in the lane, not the deviation across the lane. This is one of the main bases for the present application to achieve safety and reliability. The judgment logic can be fine-tuned according to the requirements of the OEMs. For example, when the lane width meets the threshold and there is a road edge on the right side (curb, guardrail, temporary road repair barrels, etc.), the system will follow the driving habits of human drivers and will no longer strictly drive in the center, but will drive away from the guardrail. For example, when there is a truck in the next lane on one side of the road where the vehicle is driving, and there is no truck on the other side, the system will also judge whether the relative speed, relative longitudinal distance and other thresholds between the vehicle and the truck meet the conditions. If so, the system will no longer strictly drive in the center, but will drive away from the truck side.

[0041] Determine whether the lane centering control needs to change the planned path at this time, no longer strictly maintain the lane centering control, but re-plan a reasonable route to let the vehicle drive along the new path, so that the vehicle can drive in the lane and have a certain safety distance from dangerous areas or targets, such as guardrails, large trucks, etc., so as to provide a better functional experience. This strategy is the active lane departure control strategy of this application. ;

[0042] In the re-planning of the tracking path, in order to ensure the stability and safety of the vehicle and the rationality of the path:

[0043] 1) Construct a vehicle dynamics model based on the vehicle lateral dynamics y0=Cx, x is the state space variable, y is the lateral error, Ψ is the heading error; A is the system matrix; B is the control matrix; C is the output matrix; y0 is the output variable (selected according to the research problem, for example, y0=x can be selected here, then C is the unit matrix); δ is the front wheel steering angle.

[0044] 2) Model predictive control (MPC) is a control strategy based on a prediction model. In the entire rolling optimization process, MPC generally has three key steps, namely, prediction model, rolling optimization, and feedback correction. Based on these three key steps, the basic principles of MPC are as follows: Figure 2 shown.

[0045] The entire control process always has an expected reference output, as shown in curve 1 (reference trajectory -1) in the figure. Therefore, at time k in the figure (the current time), the model predictive control algorithm will predict the future period (prediction time domain) [k, k+N] based on the current measurement value and prediction model. p ] system output, as shown in the curve 2 (predicted output -2) in the figure. Here, by optimizing and solving a certain objective function that satisfies some constraints, the control domain [k, k+N c ], and the control increment outside the control time domain remains 0, as shown in curve 4 in the figure, but only one element of this set of output sequences (such as the first one, or the minimum value, etc.) is used as the actual control quantity at that moment. Then when the system reaches the next moment k+1, the above solution process is repeated, and the optimization problem at each moment is completed as time goes by, so as to realize the control of the controlled object and achieve the control purpose).

[0046] In order to obtain the vehicle prediction model, according to the vehicle dynamics model in 1), the vehicle prediction model Y(k)=Ψ is constructed k ξ(k)+Θ k ΔU(k); Y(k) is the output variable sequence in the prediction time domain [yo(k+1), yo(k+2),…yo(k_Np)]; Ψk is the system matrix of the prediction model; ξ(k) is the new state quantity composed of the current state quantity and the input control quantity of the previous moment [x(k), δ(k-1)]; θk is the prediction model control matrix; ΔU(k) ​​is the prediction input increment sequence, [Δδ(k), Δδ(k+1)…Δδ(k+Nc)]

[0047] 3) In order to keep the vehicle away from potential dangerous targets such as guardrails or large trucks next to the vehicle, the lane area is demarcated as unfeasible, subfeasible, and feasible according to the position of the guardrail or large truck (see Figure 1 ), the infeasible area is the area outside the lane, which will build a hard constraint for the system. This area indicates that the software cannot plan a trajectory through this area, and the software needs to limit the vehicle control from running out of the lane;

[0048] The right side of this lane is the edge of the road or there is a large truck in the next lane. The area in front and behind the large truck is within a certain range (10m in front and behind + the value that increases dynamically according to the speed). The half-road width on the same side is taken as the sub-feasible area. This area means that the vehicle can pass through the area in the center, but this area is not the optimal driving route. For example, if there is a large truck next to it, the vehicle passes through the center through this area, but the driver will not feel so good. The driver may feel that he is closer to the truck, because generally when passing a truck, the driver will subconsciously deviate a little bit to the side away from the truck. In order to better describe the sub-feasible area in the future, it is necessary to introduce the penalty function of the sub-feasible area. In this way, in the subsequent control planning, when the vehicle can pass through the center and deviate to one side at the same time, because the penalty function of passing through the sub-feasible area is higher, it will eventually choose to deviate to one side.

[0049] The obstacle avoidance penalty function is adjusted by the distance between the vehicle and the sub-feasible area. The closer the vehicle is to the sub-feasible area, the larger the obstacle avoidance penalty function is; the farther the vehicle is from the obstacle, the smaller the obstacle avoidance penalty function is. Select the following obstacle avoidance function expression (this expression is the penalty function for the vehicle to a certain point)

[0050]

[0051] in

[0052] K ob ——Obstacle avoidance penalty function weight coefficient;

[0053] v——vehicle speed;

[0054] x, y——the horizontal and vertical coordinates of the vehicle's center of mass in the inertial coordinate system;

[0055] x ob ,y ob ——The horizontal and vertical coordinates of the obstacle point in the inertial coordinate system;

[0056] λ——A very small positive number to prevent the denominator from being equal to 0

[0057] However, there are countless points in the sub-feasible area, so in practice the area will be divided into regions and points will be taken. For example, points will be taken at intervals of 1m in the horizontal and vertical directions of the sub-feasible area (first divide along the edge, and if the area is wide enough, then divide internally, and finally points less than 1m will be taken on the boundary). The sub-feasible area will be divided into n reasonable points, and the position penalty function of the vehicle to each point will be calculated as one of the evaluation indicators of the cost function of the path planning system. In step 2), the position information and motion information of the predicted time domain will be obtained, and in step 3), the area of ​​the sub-feasible area division points will be obtained. According to the above single-point penalty function, the overall penalty for the entire predicted time domain trajectory can be obtained. Penalty function x, y are the position information in the prediction time domain of the prediction model; x(ob,i), y(ob,i) are the position information of the sub-feasible area division point, vx vy are the vehicle speed information in the prediction time domain of the prediction model, and k(ob,i) is the weight of the sub-feasible area division point. Generally speaking, the closer to the lane line in the horizontal direction, the larger the coefficient, and the closer to the current vehicle position in the vertical direction, the larger the weight. This weight function is mainly used for the calculation of step 4). In theory, it ensures that the vehicle is centered and does not deviate from the lane. The vehicle can drive in the center. After the penalty function of the sub-feasible area is introduced, the cost function of the sub-feasible area will be larger, so as to achieve driving that is not completely centered and slightly deviates from the sub-feasible area.

[0058] 4) Considering that the vehicle must travel within the lane, there will be a hard constraint on the lateral position of the vehicle, y_thr>|y|, y is the lateral position of the vehicle corresponding to the prediction model, y_thr is the lateral position of the vehicle corresponding to the vehicle crossing the line. According to step 2), |Y(k)|=|[yo(k+1),yo(k+2),…yo(k_Np)]|<[y_thr,y_thr…,y_thr] is obtained, thus obtaining the first hard constraint of the prediction model;

[0059] In order to ensure vehicle stability, vehicle dynamics safety constraints will be considered in path planning, such as constraints on the front wheel turning angle. Related research shows that the following constraints can be given for the description of vehicle lateral stability, |δ|<3.5deg; considering the comfort of the vehicle and the passing line, the constraint on the vehicle lateral acceleration, |ay_max|<1.5m / s^2; for the constraint on the vehicle side slip angle, for vehicle stability research, the following constraints can be set, which are the constraints on the vehicle center of mass side slip angle β on good roads and the constraints on the vehicle center of mass side slip angle β on ice and snow roads, respectively, -12°<β<12°or-2°<β<2°. At present, we only consider normal roads, so we choose the first constraint. According to steps 1) and 2), it can be seen that this constraint is a constraint on the first state of ξ(k), |ξ(k)*[1 0]'|<12deg. And finally, these vehicle dynamics safety constraints are converted into constraints on the state variables and control variables of the vehicle prediction model constructed in 2);

[0060] 5) Based on the contents from 2) to 4), the y value of the vehicle in the center o_refThat is, the center line of the lane line. Considering each prediction time domain, the centering performance is the best, the control amount is the smallest, the change of the control amount is the smallest, the penalty function is the smallest, and try to drive along the center line of the lane. Substitute the prediction information in the prediction time domain of the model prediction equation obtained in 2) and the cost function obtained in the sub-feasible area obtained in step 3) (Note: If there is no sub-feasible area divided by the curb, the cost function of the sub-feasible area is 0, and only the centering performance needs to be considered). Propose the corresponding optimization objective function ;

[0062] Nc – control horizon, i.e. the number of control steps in the prediction model;

[0063] Nc-1 is Nc minus 1;

[0064] J-cost function, the goal is to find the minimum value of the function in the prediction time domain;

[0065] N_p – prediction time domain, that is, how many steps to predict;

[0066] y o – vehicle kinematic output variables, The state quantity in step 1, the most important one calculated by the equation is the lateral position state quantity;

[0067] y o_ref – Vehicle status ref value, y_ref=0 means the vehicle is centered;

[0068] Δu – control input increment;

[0069] u – control input quantity;

[0070] J ob – Substitute the cost function of the sub-feasible area obtained in step 3) (Note: if there is no sub-feasible area divided by the roadside, the cost function of the sub-feasible area is 0, and only the centering performance needs to be considered);

[0071] ρε 2 – Adding a relaxation factor with a small value ensures that the system can obtain a suboptimal solution when there is no optimal solution, thus avoiding the situation where there is no feasible solution;

[0072] Q, R, S – weight matrices;

[0073] The constraints of the problem are constructed by combining the hard constraints of the vehicle position and the vehicle stability and safety constraints. By combining the objective function and the corresponding hard constraints, the solution of the objective function problem can be converted into the solution of the QP problem, so as to solve the control sequence that satisfies the avoidance of dangerous areas while ensuring the stability and safety of the vehicle. Through rolling optimization calculations, the vehicle can finally avoid the potential dangerous area of ​​the current road while ensuring the stability and safety of the vehicle.

[0074] Obviously, the specific implementation of the present invention is not limited to the above-mentioned methods. As long as various non-substantial improvements are made using the method concept and technical solution of the present invention, they are all within the protection scope of the present invention.

Claims

1. A lane keeping control method based on road conditions, characterized in that: include, Real-time detection of lane line information and information on both sides of the lane line; Based on the detected lane line information and the information on both sides of the lane line, it is determined whether the lane line keeping is controlled according to the lane centering control strategy; if so, the lane centering control strategy is used to control the vehicle, otherwise, the active deviation strategy is used to control the vehicle; The active deviation strategy refers to controlling the vehicle to travel within the lane and deviate to both sides of the lane.

2. The lane keeping control method based on road conditions as claimed in claim 1, characterized in that: Based on the information on both sides of the lane line, the obstacles and vehicle information on both sides of the lane are identified, and the vehicle is controlled to drive within the lane and deviate to the side of the lane away from obstacles or vehicles in the active deviation strategy.

3. The lane keeping control method based on road conditions as claimed in claim 1, characterized in that: In the active deviation strategy, the vehicle's driving path is planned based on the real-time detected lane line information and the information on both sides of the lane line, and the vehicle is controlled according to the driving path.

4. The lane keeping control method based on road conditions as claimed in claim 4, characterized in that: Planning the vehicle's driving path in the active departure strategy includes: S1. Establish vehicle dynamics model; S2. Establishing a vehicle prediction model based on the vehicle dynamics model; S3: Divide the area in the lane based on the information on both sides of the detected lane line and establish hard constraints on the route in the lane S4. Establish hard constraints on the lateral position in the lane and dynamic safety constraints; S5. Solve the pre-established objective function based on the constraints in steps S3 and S4 to obtain corresponding path control parameters.

5. The lane keeping control method based on road conditions as claimed in claim 4, characterized in that: The dynamic safety constraints in S4 include constraints on the front wheel turning angle, vehicle lateral acceleration, and vehicle sideslip angle.

6. The lane keeping control method based on road conditions as claimed in claim 4, characterized in that: The objective function established is .

7. The lane keeping control method based on road conditions according to any one of claims 1 to 6, characterized in that: The hard constraints on the vehicle position and the vehicle stability and safety constraints are established as constraint conditions. Combined with the objective function and the corresponding hard constraints, the solution to the objective function problem is converted into the solution to the QP problem, so as to solve the control sequence that satisfies the avoidance of dangerous areas while ensuring the stability and safety of the vehicle. Through rolling optimization calculations, the vehicle can finally avoid the potential dangerous area of ​​the current road while ensuring the stability and safety of the vehicle, and output the corresponding control parameter sequence information.

8. The lane keeping control method based on road conditions according to any one of claims 1 to 6, characterized in that: The hard constraints of a vehicle refer to the position constraints of the vehicle in its own lane; The dynamic safety constraints of a vehicle refer to the conditions satisfied by the vehicle's front wheel turning angle, vehicle lateral acceleration, and vehicle sideslip angle while satisfying the vehicle's lateral stability and passability.

9. The lane keeping control method based on road conditions according to any one of claims 1 to 6, characterized in that: In step S3, the hard constraints of the route within the lane include dividing the area within the lane into infeasible, subfeasible, and feasible areas, and the division criteria are obstacles on both sides of the lane, vehicle types, and road edge information.

Images