Automatic driving vehicle risk avoiding decision-making method based on real-time risk assessment

Abstract

The invention provides an automatic driving vehicle risk avoiding decision-making method based on real-time risk assessment, which comprises the following steps of: firstly, acquiring environment information around an automatic driving vehicle through various devices, and assessing the real-time operation risk of the automatic driving vehicle based on the acquired data; judging whether the automatic driving vehicle needs to make a risk avoiding decision or not according to the risk threshold value; the method comprises the following steps: selecting an automatic driving vehicle risk avoiding decision-making method, namely a way-giving risk avoiding decision-making method or a bypass risk avoiding decision-making method, by combining the state of a high-risk traffic element, and generating a risk avoiding decision-making scheme through a path-speed separation trajectory planning method: when the way-giving risk avoiding decision-making method is selected, the automatic driving vehicle adopts an original trajectory planning path; when a detour risk avoiding decision-making method is selected, a new detour path is generated through a control point selection method based on a five-order Bezier curve; the acceleration is calculated in combination with the risk assessment result, a speed control scheme is obtained, dynamic risk avoiding trajectory planning of the autonomous vehicle is completed, and the operation safety of the autonomous vehicle is improved.

Description

Technical Field

[0001] The present invention relates to the fields of autonomous driving, safe driving and trajectory planning, and in particular to a risk avoidance decision-making method for an autonomous driving vehicle based on real-time risk assessment. Background Art

[0002] With the rapid development of the automobile industry and the continuous advancement of intelligent driving technology, autonomous vehicles have become an important development direction in the future transportation field. Autonomous vehicles have significantly improved the safety performance and driving convenience of vehicles by integrating multiple technologies such as artificial intelligence, big data, and sensors. However, in a complex traffic environment, how to ensure that autonomous vehicles can make risk avoidance decisions quickly and accurately when facing sudden dangers is still a key technical problem that needs to be solved urgently.

[0003] At present, the research methods related to risk avoidance decision-making for autonomous vehicles can be roughly divided into two categories. One is based on advanced sensor systems and data processing technologies, and predicts and evaluates potential accident risks by real-time monitoring and analysis of environmental information around the vehicle. For example, the elements in the hypothetical scenario such as SATTEL will generate field strength, and the risk of the main vehicle is evaluated based on the repulsive force generated by the field strength acting on the vehicle. YANG et al. used the relative distance between the vehicle and the obstacle as the core parameter for field strength calculation to evaluate vehicle risk. This type of research can dynamically evaluate the vehicle operation risk in real time and generate risk heat maps to assist drivers in driving, but there is still a lot of room for further vehicle risk avoidance trajectory planning in terms of how to use risk assessment results. The other type is to incorporate safety into trajectory planning as a constraint, and to understand and identify complex traffic scenes through rule-based or machine learning-based methods to complete safe trajectory planning. For example, Ammour et al. used model predictive control methods to generate constraints based on traffic conditions, allowing autonomous vehicles to make efficient decisions on lane keeping, lane transfer, overtaking or braking to avoid collisions with road participants. Shin et al. designed a vehicle steering avoidance method based on the driver's braking model for small overlap collision scenarios, considering the risk of collision with the following vehicle during emergency braking. Fu et al. fully considered efficiency, accuracy and passenger comfort, and proposed an autonomous braking strategy based on deep reinforcement learning in emergency situations. This type of method can generate safe trajectories for autonomous driving vehicles, but the risk factors are not considered enough, and the generated trajectory may not meet the expectations of human risk avoidance decisions, which hinders people's trust and acceptance of the popularization of autonomous driving technology.

[0004] Therefore, in response to the shortcomings of existing risk avoidance decision-making methods for autonomous driving vehicles, this paper proposes a risk avoidance decision-making method for autonomous driving vehicles based on real-time risk assessment, which completes the whole process from real-time operation risk assessment of autonomous driving vehicles to dynamic planning of risk avoidance trajectories. Different from other risk avoidance decision-making methods for autonomous driving vehicles, this method performs risk avoidance trajectory planning through unified risk description, trajectory planning with path-speed separation, and path generation method based on fifth-order Bezier curves. It can be applied to various driving scenarios such as straight driving and turning, and can achieve more intelligent, safe, and reliable risk avoidance decisions. Summary of the invention

[0005] In view of the shortcomings of existing autonomous driving vehicle risk avoidance methods, the invention of this paper aims to provide a path-speed separation yield or detour risk avoidance strategy for vehicles in high-risk scenarios based on the real-time dynamic vehicle operation risk assessment results. To achieve the above purpose, this paper provides an autonomous driving vehicle risk avoidance decision-making method based on real-time risk assessment. The specific implementation steps of this method are as follows:

[0006] Step 1: Obtain the vehicle's surrounding environment information through the autonomous driving vehicle's onboard camera, radar and GPS data, including obstacles and movement information data of surrounding vehicles.

[0007] Step 2: Calculate the real-time operation risk of the autonomous driving vehicle i based on the data collected in step 1. The specific implementation steps of step 2 are as follows:

[0008] Step 21: establish a right-handed rectangular coordinate system with the autonomous driving vehicle i as the origin and the target vehicle speed direction as the positive direction of the x-axis. The risk calculation formula for other traffic elements j to the autonomous driving vehicle i at time t is:

[0009]

[0010] Among them, x i (t), y i (t) is the horizontal and vertical coordinates of the autonomous driving vehicle i at time t, x j (t), y j (t) is the horizontal and vertical coordinates of other traffic elements j at time t; R i M is the influencing factor of the road conditions of the autonomous driving vehicle i; i is the equivalent mass of the autonomous driving vehicle i; θ ij (t) represents the angle between other traffic elements and the positive direction of the x-axis at time t; v ij (t) is the relative speed between the autonomous driving vehicle i and other traffic elements j at time t.

[0011] Step 22: Take the maximum risk of other traffic factors to the autonomous driving vehicle i as the risk faced by the autonomous driving vehicle i, and the traffic factor that produces the maximum risk is recorded as j 0 , the risk faced by autonomous driving vehicle i at time t is:

[0012] Risk i (t)=max(Risk ij (t))

[0013] Step 3: Based on the risk value of the autonomous driving vehicle i obtained in step 2, determine whether a risk avoidance decision is needed; when the autonomous driving vehicle i has a risk level of Risk i Less than or equal to the risk threshold R s , driving according to the original trajectory planning; when the autonomous driving vehicle i runs at a risk level Risk i Greater than the risk threshold R s , it is necessary to make risk avoidance decisions and re-plan the trajectory of the autonomous driving vehicle i.

[0014] Step 4: Based on the comparison results obtained in step 3, a path-speed separation risk avoidance trajectory planning is provided for the autonomous driving vehicle i that needs to make a risk avoidance decision; 0 The movement state of traffic element j is selected to avoid danger by detouring or yielding: when the traffic element j 0 Risk to autonomous vehicles i i If it does not change, then adopt a detour hedging strategy; otherwise, adopt a give way hedging strategy.

[0015] Step 5: Calculate the trajectory of the yielding risk avoidance strategy. Based on the original path L0 of the autonomous driving vehicle i remaining unchanged, the speed is controlled by acceleration to reduce the risk of the autonomous driving vehicle; the acceleration magnitude of the autonomous driving vehicle i at time t is:

[0016]

[0017] Among them, v i (t) is the speed of the autonomous driving vehicle i at time t, p i (t) is the coordinate value of the autonomous driving vehicle i on the path L0 at time t, p i，d (t+Δt) is the coordinate value at which the risk at time t+Δt is equal to the risk threshold Risk when performing uniform deceleration motion on L0.

[0018] Step 6, calculate the detour avoidance strategy trajectory; use the fifth-order Bezier curve to generate the detour path of the autonomous driving vehicle i, and use speed control to keep the risk of the autonomous driving vehicle i below the risk threshold. The specific implementation steps of step 6 are as follows:

[0019] Step 61, the fifth-order Bezier curve is determined based on six control points; P1 is the starting point of trajectory planning, that is, the position coordinate of the autonomous driving vehicle i at time t; P2 is the starting speed direction control point, which is the position coordinate of point P1 moving 0.1m in the speed direction of the autonomous driving vehicle i; points P3+, P4+ (P3-, P4-) are the control points for detours in the clockwise (counterclockwise) direction; P6 is the end point of trajectory planning, that is, the end position coordinate of the autonomous driving vehicle i on the original path; P5 is the end speed direction control point, which is the position coordinate of point P6 moving 0.1m in the opposite direction of the speed of the autonomous driving vehicle i.

[0020] Step 62, solve the position coordinates of the control points P3+, P4+ (P3-, P4-). 0 The geometric center of is the center of the circle, the radius of its circumscribed circle is R, and circle C is drawn with R as the radius j1 , through P1, rotate clockwise along P1 and P2 to make circle C j 1 tangent, tangent point is P3+, through P1 and rotate counterclockwise along P1 and P2 to make circle C j Tangent, the tangent point is P3-; through P6, rotate counterclockwise along P6 and P5 to make a circle tangent point is P4+, through P6, rotate clockwise along P6 and P5 to make circle C j Tangent line, tangent point is P4-.

[0021] Step 63, generate a detour path L1 according to the control point coordinates P1, P2, P3+, P4+, P5, P6 and the Bezier curve formula, generate a detour path L2 according to the control point coordinates P1, P2, P3-, P4-, P5, P6 and the Bezier curve formula, and determine the coordinates of the path point i of the autonomous driving vehicle as follows:

[0022]

[0023] Step 64, check the generated paths L1, L2 and traffic element j 0 Is there an intersection? If path L1 has an intersection, draw a circle C with R1+Δr as radius. j1 Repeat steps 62 to 64 until the path L1 and the traffic element j 0 If there is no intersection, then path L1 is a detour path for the autonomous vehicle i; if path L2 has an intersection, then a circle C is drawn with R2+Δr as the radius. j2 Repeat steps 62 to 64 until the path L2 and the traffic element j 0 If there is no intersection, path L2 is a detour path for the autonomous driving vehicle i; Δr is the change in the radius of the circle each time.

[0024] Step 65, compare the lengths of paths L1 and L2, and the path with the shorter length is the final detour path L of the autonomous driving vehicle i.

[0025] Step 66, based on the path L, adopt the same speed control method as the yielding risk avoidance strategy, i.e., step 5, to generate the final detour risk avoidance strategy trajectory of the autonomous driving vehicle i.

[0026] Step 7, at time t, the autonomous driving vehicle i executes trajectory planning for risk avoidance decision-making. At time t+1, it is determined again according to steps 1-3 whether a risk avoidance decision is needed. If necessary, steps 4 to 6 are repeated to generate a new risk avoidance decision plan. By performing a decision cycle at each moment, the safe driving of the autonomous driving vehicle can be ensured. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a block diagram of the overall concept of the present invention;

[0028] Figure 2 Diagram of the control point selection for the fifth-order Bezier curve. DETAILED DESCRIPTION

[0029] The present invention is described in detail below with reference to the accompanying drawings and implementation schemes. It should be understood that this example is only used to illustrate the present invention and is not used to limit the scope of the present invention. The overall idea of ​​the automatic driving vehicle risk avoidance decision method based on real-time risk assessment provided by the present invention is as follows: Figure 1 As shown, the specific implementation method includes the following steps:

[0030] Step 1: Obtain the vehicle's surrounding environment information through the autonomous driving vehicle's onboard camera, radar and GPS data, including obstacles and movement information data of surrounding vehicles.

[0031] Step 2: Calculate the real-time operation risk of the autonomous driving vehicle i based on the data collected in step 1. The specific implementation steps of step 2 are as follows:

[0032] Step 21: establish a right-handed rectangular coordinate system with the autonomous driving vehicle i as the origin and the target vehicle speed direction as the positive direction of the x-axis. The risk calculation formula for other traffic elements j to the autonomous driving vehicle i at time t is:

[0033]

[0034] Among them, x i (t), y i (t) is the horizontal and vertical coordinates of the autonomous driving vehicle i at time t, x j (t), y j (t) is the horizontal and vertical coordinates of other traffic elements j at time t; R i M is the influencing factor of the road conditions of the autonomous driving vehicle i; i is the equivalent mass of the autonomous driving vehicle i; θ ij(t) represents the angle between other traffic elements j and the positive direction of the x-axis at time t; v ij (t) is the relative speed between the autonomous driving vehicle i and other traffic elements j at time t.

[0035] Step 22: Take the maximum risk of other traffic factors to the autonomous driving vehicle i as the risk faced by the autonomous driving vehicle i, and the traffic factor that produces the maximum risk is recorded as j 0 , the risk faced by autonomous driving vehicle i at time t is:

[0036] Risk i (t)=max(Risk ij (t))

[0037] Step 3: Based on the risk value of the autonomous driving vehicle i obtained in step 2, determine whether a risk avoidance decision is needed; when the autonomous driving vehicle i has a risk level of Risk i Less than or equal to the risk threshold R s , driving according to the original trajectory plan; when the autonomous driving vehicle operation risk level Risk i Greater than the risk threshold R s , it is necessary to make risk avoidance decisions and re-plan the trajectory of the autonomous driving vehicle i. Risk threshold R s Calibrated according to the actual traffic environment and safety requirement level.

[0038] Step 4: Based on the comparison results obtained in step 3, a path-speed separation risk avoidance trajectory planning is provided for the autonomous driving vehicle i that needs to make a risk avoidance decision; 0 The movement state of traffic element j is selected to avoid danger by detouring or yielding: when the traffic element j 0 Risk to autonomous vehicles i i If the speed limit is unchanged, then the detour avoidance strategy is adopted. Otherwise, the yielding avoidance strategy is adopted. Δt is determined according to the road speed limit. The higher the road speed limit, the smaller the time interval Δt is.

[0039] Step 5: Calculate the trajectory of the yielding risk avoidance strategy. Based on the original path L0 of the autonomous driving vehicle i remaining unchanged, the speed is controlled by acceleration to reduce the risk of the autonomous driving vehicle; the acceleration magnitude of the autonomous driving vehicle i at time t is:

[0040]

[0041] Among them, v i (t) is the speed of the autonomous driving vehicle i at time t, p i (t) is the coordinate value of the autonomous driving vehicle i on the path L0 at time t, p i，d(t+Δt) is the coordinate value at which the risk at time t+Δt is equal to the risk threshold Risk when performing uniform deceleration motion on L0.

[0042] Step 6, calculate the detour avoidance strategy trajectory; use the fifth-order Bezier curve to generate the detour path of the autonomous driving vehicle i, and use speed control to keep the risk of the autonomous driving vehicle i below the risk threshold. The specific implementation steps of step 6 are as follows:

[0043] Step 61, the fifth-order Bezier curve is determined based on six control points, and the selection rule is as follows: Figure 2 As shown. P1 is the starting point of trajectory planning, that is, the position coordinate of the autonomous driving vehicle i at time t; P2 is the starting speed direction control point, which is the position coordinate of point P1 moving 0.1m in the direction of the speed of the autonomous driving vehicle i; points P3+, P4+ (P3-, P4-) are the control points for detours in the clockwise (counterclockwise) direction; P6 is the end point of trajectory planning, that is, the end position coordinate of the autonomous driving vehicle i on the original path; P5 is the end speed direction control point, which is the position coordinate of point P6 moving 0.1m in the opposite direction of the speed of the autonomous driving vehicle i.

[0044] Step 62, solve the position coordinates of the control points P3+, P4+ (P3-, P4-). 0 The geometric center of is the center of the circle, the radius of its circumscribed circle is R, and circle C is drawn with R as the radius j1 , through P1, rotate clockwise along P1 and P2 to make circle C j 1 tangent, tangent point is P3+, through P1 and rotate counterclockwise along P1 and P2 to make circle C j Tangent, the tangent point is P3-; through P6, rotate counterclockwise along P6 and P5 to make a circle tangent point is P4+, through P6, rotate clockwise along P6 and P5 to make circle C j Tangent line, tangent point is P4-.

[0045] Step 63, generate a detour path L1 according to the control point coordinates P1, P2, P3+, P4+, P5, P6 and the Bezier curve formula, generate a detour path L2 according to the control point coordinates P1, P2, P3-, P4-, P5, P6 and the Bezier curve formula, and determine the coordinates of the path point i of the autonomous driving vehicle as follows:

[0046]

[0047] Step 64, check the generated paths L1, L2 and traffic element j 0 Is there an intersection? If path L1 has an intersection, draw a circle C with R1+Δr as radius. j1 Repeat steps 62 to 64 until the path L1 and the traffic element j 0If there is no intersection, then path L1 is a detour path for the autonomous vehicle i; if path L2 has an intersection, then a circle C is drawn with R2+Δr as the radius. j2 Repeat steps 62 to 64 until the path L2 and the traffic element j 0 If there is no intersection, path L2 is a detour path for the autonomous driving vehicle i. Δr is the change in the radius of the circle each time, which is determined by the computing resources. The smaller the value of Δr, the higher the space utilization rate of the detour trajectory. The larger the value of Δr, the fewer the calculation times.

[0048] Step 65, compare the lengths of paths L1 and L2, and the path with the shorter length is the final detour path L of the autonomous driving vehicle i.

[0049] Step 66, based on the path L, adopt the same speed control method as the yielding risk avoidance strategy, i.e., step 5, to generate the final detour risk avoidance strategy trajectory of the autonomous driving vehicle i.

[0050] Step 7, at time t, the autonomous driving vehicle i executes trajectory planning for risk avoidance decision-making. At time t+1, it is determined again according to steps 1-3 whether a risk avoidance decision is needed. If necessary, steps 4 to 6 are repeated to generate a new risk avoidance decision plan. By performing a decision cycle at each moment, the safe driving of the autonomous driving vehicle can be ensured.

[0051] The above steps describe the implementation process of the present invention in detail, but the present invention is not limited to the specific details of the above implementation. Anything within the scope of the present invention should not be excluded from the protection scope of the present invention.

Claims

1. A risk avoidance decision-making method for an autonomous driving vehicle based on real-time risk assessment, characterized in that: The following steps are involved: Step 1: Obtaining the vehicle's surrounding environment information through the autonomous driving vehicle's onboard camera, radar, and GPS data, including obstacles and movement information data of surrounding vehicles; Step 2: Calculate the real-time operation risk of the autonomous driving vehicle i based on the data collected in step 1. The specific implementation steps of step 2 are as follows: Step 21, establish a right-hand plane rectangular coordinate system with the autonomous driving vehicle i as the origin and the speed direction of the target vehicle as the positive direction of the x-axis. The real-time risk calculation formula of other traffic elements j to the autonomous driving vehicle i at time t is: Among them, x i (t), y i (t) is the horizontal and vertical coordinates of the autonomous driving vehicle i at time t, x j (t), y j (t) is the horizontal and vertical coordinates of other traffic elements j at time t; R i M is the influencing factor of the road conditions of the autonomous driving vehicle i; i is the equivalent mass of the autonomous driving vehicle i; θ ij (t) represents the angle between other traffic elements j and the positive direction of the x-axis at time t; v ij (t) is the relative speed between the autonomous driving vehicle i and other traffic elements j at time t; Step 22: Take the maximum risk of other traffic factors to the autonomous driving vehicle i as the risk faced by the autonomous driving vehicle i. The traffic factor that generates the maximum risk is recorded as j0. The risk faced by the autonomous driving vehicle i at time t is: Risk i (t)=max(Risk ij (t)) Step 3: Based on the risk value of the autonomous driving vehicle i obtained in step 2, determine whether a risk avoidance decision is needed; when the autonomous driving vehicle i has a risk level of Risk i Less than or equal to the risk threshold R s , driving according to the original trajectory planning; when the autonomous driving vehicle i runs at a risk level Risk i Greater than the risk threshold R s , it is necessary to make risk avoidance decisions and re-plan the trajectory of the autonomous driving vehicle i; Step 4: Based on the comparison results obtained in step 3, provide a path-speed separation risk avoidance trajectory planning for the autonomous driving vehicle i that needs to make a risk avoidance decision; select a detour risk avoidance strategy or a yield risk avoidance strategy according to the motion state of the traffic element j0: when the risk Risk generated by the traffic element j0 to the autonomous driving vehicle i within the Δt time interval is i If it does not change, then adopt the strategy of bypassing and avoiding risks. Otherwise, adopt the strategy of giving way and avoiding risks. Step 5: Calculate the trajectory of the yielding risk avoidance strategy. Based on the original path L0 of the autonomous driving vehicle i remaining unchanged, the speed is controlled by acceleration to reduce the risk of the autonomous driving vehicle; the acceleration magnitude of the autonomous driving vehicle i at time t is: Among them, v i (t) is the speed of the autonomous driving vehicle i at time t, p i (t) is the coordinate value of the autonomous driving vehicle i on the path L0 at time t, p i,d (t+Δt) is the coordinate value at which the risk at time t+Δt is equal to the risk threshold Risk when performing uniform deceleration motion on L0; Step 6, calculate the detour avoidance strategy trajectory; use the fifth-order Bezier curve to generate the detour path of the autonomous driving vehicle i, and use speed control to keep the risk of the autonomous driving vehicle i below the risk threshold; the specific implementation steps of step 6 are as follows: Step 61, the fifth-order Bezier curve is determined based on six control points; P1 is the starting point of trajectory planning, that is, the position coordinate of the autonomous driving vehicle i at time t; P2 is the starting speed direction control point, which is the position coordinate of point P1 moving 0.1m in the speed direction of the autonomous driving vehicle i; points P3+, P4+ (P3-, P4-) are the control points for detours in the clockwise (counterclockwise) direction; P6 is the end point of trajectory planning, that is, the end point position coordinate of the autonomous driving vehicle i on the original path; P5 is the end speed direction control point, which is the position coordinate of point P6 moving 0.1m in the opposite direction of the speed of the autonomous driving vehicle i; Step 62, solve the position coordinates of the control points P3+, P4+ (P3-, P4-). Take the geometric center of the traffic element j0 as the center of the circle, and the radius of the circumscribed circle is R. Draw a circle C with R as the radius. j1 , through P1, rotate clockwise along P1 and P2 to make circle C j 1 tangent, tangent point is P3+, through P1 and rotate counterclockwise along P1 and P2 to make circle C j Tangent, the tangent point is P3-; through P6, rotate counterclockwise along P6 and P5 to make a circle tangent point is P4+, through P6, rotate clockwise along P6 and P5 to make circle C j Tangent line, tangent point is P4-; Step 63, generate a detour path L1 according to the control point coordinates P1, P2, P3+, P4+, P5, P6 and the Bezier curve formula, generate a detour path L2 according to the control point coordinates P1, P2, P3-, P4-, P5, P6 and the Bezier curve formula, and determine the coordinates of the path point i of the autonomous driving vehicle as follows: Step 64, check whether the generated paths L1, L2 and the traffic element j0 have intersections. If path L1 has an intersection, draw a circle C with a radius of R1+Δr. j1 Repeat steps 62 to 64 until the path L1 has no intersection with the traffic element j0. Then the path L1 is a detour path for the autonomous driving vehicle i. If the path L2 has an intersection, draw a circle C with a radius of R2+Δr. j2 , repeat steps 62 to 64 until the path L2 has no intersection with the traffic element j0, then the path L2 is a detour path for the autonomous driving vehicle i; Δr is the change in the radius of the circle each time; Step 65, comparing the lengths of paths L1 and L2, the path with the shorter length is the final detour path L of the autonomous driving vehicle i; Step 66, based on the path L, adopt the same speed control method as the yielding risk avoidance strategy, i.e., step 5, to generate the final detour risk avoidance strategy trajectory of the autonomous driving vehicle i; Step 7, at time t, the autonomous driving vehicle i executes trajectory planning for risk avoidance decision-making. At time t+1, it is determined again according to steps 1-3 whether a risk avoidance decision is needed. If necessary, steps 4 to 6 are repeated to generate a new risk avoidance decision plan. By performing a decision cycle at each moment, the safe driving of the autonomous driving vehicle can be ensured.

2. The method according to claim 1, characterized in that This method proposes a universal risk avoidance decision-making method that can realize the whole process from real-time operation risk assessment of autonomous driving vehicles to dynamic planning of risk avoidance trajectories through unified risk description, path-speed separation trajectory planning, and path generation method based on fifth-order Bezier curves, thereby helping autonomous driving vehicles to complete safer trajectory planning.

Images