Method for predicting movement of surrounding vehicle in surrounding area of vehicle

Abstract

The invention relates to a method for predicting a movement (B) of a surrounding vehicle (2) in a surrounding area of a vehicle (1), in which a front lateral velocity (vQ) at a lateral front position of the surrounding vehicle (2) is determined on the basis of sensor data detected by means of an on-board sensor system (3) of the vehicle (1) in order to determine a rotational movement (DB) of the surrounding vehicle (2), characterized in that the front lateral velocity (vQ) at a lateral front position of the surrounding vehicle (2) is determined on the basis of the sensor data detected by means of the on-board sensor system (3) of the vehicle (1). When determining, on the basis of the sensor data, whether a surrounding vehicle (2) traveling in the same direction will cut in front of the vehicle (1),-determining, on the basis of the sensor data, a lane route of a lane (FS1) in which the vehicle (1) is located, a yaw angle (psi) as an angle between the lane route and the direction of the surrounding vehicle (2), and a traveling speed (v) of the surrounding vehicle (2); defining a pivot point (DP) of the surrounding vehicle (2), and predicting a parking position (A1 to An) of the surrounding vehicle (2) for a plurality of future time steps by means of a single lane model (7) on the basis of the front lateral velocity (vQ), the yaw angle (psi), the travel speed (v) and the pivot point (DP) of the surrounding vehicle (2). The invention also relates to a method for operating an autonomous vehicle (1).

Description

Technical Field

[0001] This invention relates to a method for predicting the movement of surrounding vehicles in the area surrounding a vehicle, as described in the preamble of claim 1. The invention also relates to a method for operating an autonomous vehicle. Background Technology

[0002] DE 10 2013 019 621 A1 discloses a method for determining the motion of objects around a vehicle based on radar data detected by a radar sensor. To determine the motion, a position profile / distribution and a velocity profile / distribution of the object determined from the radar data are evaluated. In the velocity profile evaluation, the rotational motion of the object is determined within a time step based on the relative velocities of reflection points located at different positions on the object relative to the radar sensor. Additionally, in the velocity profile evaluation, the rotational motion of the object is determined over several consecutive time steps, from precise to a single reflection point relative to the radar sensor at different relative velocities. The detected relative velocities are compared with stored patterns, wherein the motion of the object is determined by comparison. Summary of the Invention

[0003] The purpose of this invention is to provide a new method for predicting the movement of surrounding vehicles in the area surrounding a vehicle and a new method for operating an autonomous vehicle.

[0004] According to the present invention, this objective is achieved by a method for predicting the motion of surrounding vehicles having the features of claim 1 and a method for operating an autonomous vehicle having the features of claim 9.

[0005] Advantageous designs of the present invention are the subject of the dependent claims.

[0006] In a method for predicting the motion of surrounding vehicles in the area surrounding a vehicle, in order to determine the rotational motion of the surrounding vehicles, the forward lateral velocity of the side-front position of the surrounding vehicles is determined based on sensor data detected by means of the vehicle's onboard sensor system (e.g., camera sensor system, radar sensor system, and / or lidar sensor system).

[0007] According to the present invention, the method is characterized in that, when determining, based on detected sensor data, whether surrounding vehicles traveling in the same direction will cut in front of the vehicle, - Based on sensor data, determine the lane line of the vehicle, the yaw angle (the angle between the lane line and the directions of surrounding vehicles), and the speeds of the surrounding vehicles. - Define the pivot point of the surrounding vehicles, and - Using a single-lane model, the dwell positions of surrounding vehicles at multiple future time steps are predicted based on their forward lateral velocity, yaw angle, driving speed, and pivot point.

[0008] In the operation of automated, especially partially automated, highly automated, or autonomous vehicles, appropriate responses to cutting-in vehicles, in the form of adjusting speed and / or performing evasive maneuvers, are essential. In such cases, a safe distance must be maintained from the cutting-in vehicles, and in very close cut-in operations where the distance between the vehicle and surrounding vehicles is small, evasive maneuvers may be necessary to avoid collisions and enhance traffic safety. It is equally important that the response to cutting-in vehicles is not excessive to avoid jeopardizing subsequent traffic and, where possible, to avoid impacting traffic flow.

[0009] Therefore, it is necessary to reliably detect the cutting-in operation of surrounding vehicles in the vehicle lane, reliably predict the movement of surrounding vehicles during the cutting-in operation, and adapt the automatic longitudinal and / or lateral control of the vehicle to the detected cutting-in operation.

[0010] One problem here is that it is often impossible to reliably identify the driving direction indicators of surrounding vehicles cutting in, and thus to recognize the intention to change lanes, using onboard sensor systems for environmental detection. Therefore, in some cases, a lane-changing maneuver is only recognized when significant lateral movement of the surrounding vehicles is detected. Furthermore, in lane-changing maneuvers where the distance between the vehicle and the surrounding vehicles is very small, the surrounding vehicles may only be partially within the detection range of the forward-facing onboard sensor system, which is also a problem.

[0011] This method significantly improves the prediction of the motion of surrounding vehicles, particularly long vehicles such as trucks, during cut-in operations by using a single-lane model based on the stopping positions of surrounding vehicles, including their forward lateral velocity, yaw angle, speed, and pivot point. In this case, even in cut-in operations with small distances between the vehicle and surrounding vehicles, the front-side positions of surrounding vehicles can be reliably and robustly detected using the vehicle's onboard sensor system. The improvements achieved through this method allow for optimized system responses of autonomous vehicles to cutting-in surrounding vehicles, such as avoidance maneuvers and / or braking, thereby increasing traffic safety. Inappropriate system responses can thus be effectively avoided.

[0012] In one possible design of this method, the dynamics of the cut-in operation of surrounding vehicles and the relative proximity of surrounding vehicles are determined based on prediction, and the degree of danger of the cut-in situation is determined based on the determined dynamics and proximity. This determination of the degree of danger can be performed simply and reliably, and enables the autonomous vehicle to trigger an appropriate response to the cut-in surrounding vehicles.

[0013] In another possible design of this method, the measure of the danger level of the determined cut-in situation increases as the dynamics of the cut-in operations of surrounding vehicles increase and / or the distance between surrounding vehicles decreases. Therefore, the measure of danger level can be reliably determined.

[0014] In another possible design of this method, the cutting-in operation of a surrounding vehicle is identified by the lateral movement of a point on the side-front of the surrounding vehicle relative to the vehicle. Due to the reliable and robust detection of this point by the onboard sensor system, the cutting-in operation can be identified particularly reliably based on the lateral movement of this point.

[0015] In another possible design of this method, the pivot point of the surrounding vehicles is defined on the rear axle of the surrounding vehicles, since the rotation of the surrounding vehicles typically occurs in this area.

[0016] In another possible design of this method, the prediction of the stopping position ends when the rear of surrounding vehicles has reached the center of the lane in which the vehicle is located. Once the rear of surrounding vehicles has reached the center of the lane in which the vehicle is located, the end of the cut-in operation can be reliably detected.

[0017] In another possible design of this method, the forward lateral velocity is determined by using a state observer and / or a Kalman filter with respect to the time derivative of a point located at the side-front position of surrounding vehicles, such that the forward lateral velocity has both translational and rotational components. Advantageously, due to the rotational component of the lateral velocity, especially for long surrounding vehicles such as trucks, a significant increase in the lateral component of the lateral velocity can be detected during a cut-in operation before the geometric center of gravity of the surrounding vehicles shifts. Therefore, the cut-in operation is particularly easy to identify at an early stage.

[0018] In another possible design of this method, the prediction of the stopping position is performed through forward integration of a single-lane model, wherein, for each integration step, the change in yaw angle of surrounding vehicles, the new yaw angle of surrounding vehicles, and the new position of the pivot point of surrounding vehicles are determined. Therefore, prediction can be performed in a simple and reliable manner.

[0019] In methods for operating vehicles that are automated, particularly partially automated, highly automated, or autonomously driven, according to the invention, the longitudinal and / or lateral movements of the vehicle are automatically controlled based on the stopping positions of surrounding vehicles determined in the aforementioned methods. Due to the particularly reliable prediction of the movements of surrounding vehicles during a cut-in operation, the optimized system response of the autonomous vehicle to the cutting-in surrounding vehicles can be optimized, such as avoidance maneuvers and / or braking operations, thereby increasing traffic safety.

[0020] In another design of this method, the type and intensity of changes in the current longitudinal and / or lateral movement of the vehicle in response to the cut-in situation are set according to the determined degree of danger of the cut-in situation. Therefore, inappropriate system responses can be effectively avoided, thereby minimizing harm to subsequent traffic and damage to traffic flow. Attached Figure Description

[0021] Embodiments of the present invention will now be explained in more detail with reference to the accompanying drawings.

[0022] in: Figure 1 A schematic plan view showing the cutting-in situation of the vehicle and surrounding vehicles; Figure 2 The illustration schematically shows the prediction of the motion of surrounding vehicles during a highly dynamic cut-in operation; Figure 3 The illustration schematically shows the prediction of the motion of surrounding vehicles during a cut-in operation with low dynamics. Figure 4 The illustration shows the process of detecting surrounding vehicles according to... Figure 1 A plan view showing the entry points; Figure 5 A model of the surrounding vehicles is shown schematically; Figure 6 The motion curves of different areas surrounding the vehicle are schematically shown during the cut-in operation; Figure 7 The diagram schematically illustrates the curves of lateral velocity in different areas surrounding the vehicle during the cut-in operation, and Figure 8 A block diagram of a device for predicting the movement of surrounding vehicles in the area around the vehicle is shown schematically.

[0023] Corresponding parts in all the accompanying figures are labeled with the same reference numerals. Detailed Implementation

[0024] Figure 1A plan view showing the lane-changing / cutting situation of vehicle 1 and surrounding vehicle 2 is shown. In this case, vehicle 1 is moving in lane FS1, while surrounding vehicle 2 (e.g., a truck) is moving in the adjacent lane FS2. In order to change lanes, surrounding vehicle 2 has initiated a cut-in operation from lane FS2 to vehicle 1's lane FS1.

[0025] Vehicle 1 is automated, particularly partially automated, highly automated, or autonomously driven. For this type of automated operation of Vehicle 1, appropriate responses to approaching surrounding vehicles 2, in the form of adjusting its speed v and / or performing evasive maneuvers, are essential. In such cases, a safe distance must be maintained from the approaching surrounding vehicles 2, and in very close approach maneuvers where the distance between Vehicle 1 and surrounding vehicles 2 is small, evasive maneuvers may be necessary to avoid collisions and enhance traffic safety. It is equally important that the response to the approaching surrounding vehicles 2 is not excessive so as not to jeopardize subsequent traffic and, if possible, not to disrupt traffic flow.

[0026] Therefore, it is necessary to reliably detect the cutting-in operation of surrounding vehicle 2 into lane FS1 of vehicle 1, reliably predict the motion B and rotational motion DB of surrounding vehicle 2 during the cutting-in operation, and adapt the automatic longitudinal and / or lateral control of vehicle 1 to the detected cutting-in operation.

[0027] Figure 2 The diagram illustrates the prediction of the motion B of surrounding vehicles 2 in the lateral direction y based on time t, starting from vehicle 1 during a highly dynamic cut-in operation. To determine motion B, the dwell positions A1 to An of surrounding vehicles 2 are predicted for several future time steps.

[0028] Figure 3 The diagram illustrates the prediction of the motion B of the surrounding vehicles 2 in the lateral direction y based on time t, starting from vehicle 1 during a cut-in operation with low dynamics. To determine motion B, the dwell positions A1 to An of the surrounding vehicles 2 are predicted for several future time steps.

[0029] As the dynamics of the cut-in operation increase, the danger level of the cut-in situation increases, and different levels of danger require vehicle 1 to respond differently to the cut-in operation.

[0030] Figure 4 This illustrates the detection of surrounding vehicles 2 by the onboard sensor system 3 of vehicle 1 within the detection area E, based on... Figure 1 A plan view showing the entry points. The vehicle sensor system 3 includes, for example, a camera sensor system, a radar sensor system, and / or a lidar sensor system.

[0031] To determine the motion B and rotational motion DB of the surrounding vehicle 2, at the front-side position of the surrounding vehicle 2, indicated by point P, the forward lateral velocity v is determined by referring to sensor data detected by the onboard sensor system 3 of vehicle 1. Q (exist Figure 5 (As shown in more detail below). In this case, it is advantageous that even in close-in maneuvers between vehicle 1 and surrounding vehicles 2, the front point P can be robustly detected by the forward-facing onboard sensor system 3.

[0032] Figure 5 A model of the surrounding vehicle 2 is shown.

[0033] Use state observer 4 and / or Kalman filter 5 to differentiate in time with respect to the previous lateral point P, such as Figure 8 This is shown in more detail to obtain the forward lateral velocity v of the surrounding vehicle 2 that is cutting in. Q .

[0034] Lateral velocity v Q For example, according to

[0035] Where: yF = coordinates of point P, xF = the distance from the front axle to the rear axle of the surrounding vehicle 2. v = the speed of the surrounding vehicles 2, and ψ = yaw angle.

[0036] The yaw angle ψ is the angle between the longitudinal axis of the surrounding vehicle 2 and a fixed reference line (e.g., the road route or road markings). The onboard sensor system 3, particularly the camera sensor system and the lidar sensor system, is capable of detecting the orientation of the surrounding vehicle 2. Therefore, the yaw angle ψ can be determined by understanding the road route from a digital map or by identifying road markings through the onboard sensor system 3.

[0037] It can be seen that the lateral velocity v Q It includes translational and rotational components. Advantageously, due to the rotational component, especially in cases where the surrounding vehicle 2 is long (e.g., a truck), the forward lateral velocity v can be detected during the cutting operation before the geometric center of gravity of the surrounding vehicle 2 moves laterally. Q A significant increase.

[0038] In this case, the rotational motion DB of the surrounding vehicle 2 occurs around the pivot point DP defined on the rear axle of the surrounding vehicle 2.

[0039] Figure 6The movement of different areas of the surrounding vehicle 2 according to time t is shown in the example of a cut-in operation lasting 8 seconds, at a driving speed v of the surrounding vehicle 2, for example, 30 km / h.

[0040] Motion B1 is the lateral movement of the front lateral point P, motion B2 is the lateral movement of the pivot point DP, and motion B3 is the lateral movement of the rear point of the surrounding vehicle 2. The curves of motions B1 to B3 show the temporal "protrusion" of the front lateral point P before the pivot point DP and the rear point.

[0041] Figure 7 This illustrates the lateral velocity v of different areas of the surrounding vehicle 2 over time t during, for example, a cut-in operation lasting 8 seconds. Q v Q2 v Q3 v Q4 The speed v of the surrounding vehicle 2 is, for example, 30 km / h.

[0042] Here, the lateral velocity v Q The lateral velocity v is the lateral velocity of the forward lateral point P. Q2 It is the lateral velocity of pivot point DP, the lateral velocity v Q3 It is the lateral velocity v of the rear point of the surrounding vehicle 2. Q4 This is the average lateral velocity. Lateral velocity v Q v Q2 v Q3 v Q4 The curve shows the "protrusion" of the front lateral point P before the pivot point DP and the rear point, especially continuing until the middle of the cutting operation.

[0043] Figure 8 A block diagram of a possible embodiment of a device 6 for predicting the motion B of a surrounding vehicle 2 in the area surrounding vehicle 1 is shown.

[0044] For this prediction, the forward lateral velocity v of the surrounding vehicles 2, determined from the time derivative of the forward lateral point P using state observer 4 and / or Kalman filter 5, is used. Q The data is fed into the single-lane model 7. Furthermore, the defined pivot point DP, yaw angle ψ, and the speed v of the surrounding vehicle 2 that cuts in are also fed into the single-lane model 7.

[0045] Based on the surrounding vehicle 2 moving at a speed v along its longitudinal axis and yaw angle ψ, rotating around a defined pivot point DP, and the front lateral point P moving at a lateral speed v... Q The assumption of moving to the center of lane FS1 is used to predict the dwell position of surrounding vehicle 2 at several future time steps from A1 to An by forward integral VI of single-lane model 7.

[0046] In this case, using the currently determined lateral velocity v Q Initialization is performed using the driving speed v and yaw angle ψ.

[0047] The distance xF between the front and rear axles of the surrounding vehicle 2 is usually difficult to detect by sensors, and therefore, for example, it is assumed to be a constant ratio to the length of the surrounding vehicle 2 based on structural experience.

[0048] For each integration step, the change in yaw angle ψ, the new yaw angle ψ, and the new position of pivot point DP are now determined by single-lane model 7. This integration is repeated until the rear of the surrounding vehicle 2 reaches the center of lane FS1.

[0049] For example, in Figure 2 and Figure 3 The image shows the possible predictions for the surrounding vehicle 2.

Claims

1. A method for predicting the motion (B) of surrounding vehicles (2) in a region surrounding a vehicle (1), wherein, To determine the rotational motion (DB) of the surrounding vehicle (2), the forward lateral velocity (v) at the front side position of the surrounding vehicle (2) is determined based on sensor data detected by means of the onboard sensor system (3) of the vehicle (1). Q ), The feature is that, when it is determined based on sensor data whether a surrounding vehicle (2) traveling in the same direction will cut in front of the vehicle (1), -Based on sensor data, determine the lane route of the lane (FS1) where the vehicle (1) is located, the yaw angle (ψ) which is the angle between the lane route and the direction of the surrounding vehicles (2), and the driving speed (v) of the surrounding vehicles (2). - Define the pivot point (DP) of the surrounding vehicles (2), and - Using a single-lane model (7), based on the forward lateral velocity (v) of the surrounding vehicles (2) Q The yaw angle (ψ), the driving speed (v), and the pivot point (DP) are used to predict the dwell position (A1 to An) of the surrounding vehicles (2) at multiple future time steps.

2. The method according to claim 1, Its features are, -Based on the prediction, determine the dynamics of the cut-in operation of the surrounding vehicle (2) and the relative proximity of the surrounding vehicle (2) to the vehicle (1), and - The degree of danger of the intrusion situation is determined based on the determined dynamics and the proximity.

3. The method according to claim 2, Its features are, As the dynamics of the cut-in operation of the surrounding vehicle (2) increase and / or the distance between the surrounding vehicle (2) and the vehicle (1) decreases, the measure of the danger level of the determined cut-in situation increases.

4. The method according to any one of the preceding claims, Its features are, The cutting-in operation of the surrounding vehicle (2) is identified by the lateral movement performed by a point (P) of the surrounding vehicle (2) located at the side front position of the surrounding vehicle (2) relative to the vehicle (1).

5. The method according to any one of the preceding claims, Its features are, The pivot point (DP) of the surrounding vehicle (2) is defined on the rear axle of the surrounding vehicle (2).

6. The method according to any one of the preceding claims, Its features are, The prediction of the stopping position (A1 to An) ends when the rear of the surrounding vehicle (2) has reached the center of the lane (FS1) where the vehicle (1) is located.

7. The method according to any one of the preceding claims, Its features are, The forward lateral velocity (v) is determined by using the time derivative of a point (P) located at the side-front position of the surrounding vehicle (2) with respect to a state observer (4) and / or a Kalman filter (5). Q ), causing the forward lateral velocity (v) Q It has translational and rotational components.

8. The method according to any one of the preceding claims, Its features are, The prediction of the stopping position (A1 to An) is performed by the forward integration (VI) of the single-lane model (7), wherein, for each integration step, the change of the yaw angle (ψ) of the surrounding vehicle (2), the new yaw angle (ψ) of the surrounding vehicle (2), and the new position of the pivot point (DP) of the surrounding vehicle (2) are determined.

9. A method for operating an autonomous vehicle (1), wherein, The longitudinal and / or lateral movement of the vehicle (1) is automatically controlled based on the stopping positions (A1 to An) of the surrounding vehicles (2) determined in the method of any one of the preceding claims.

10. The method according to claim 9, wherein, The type and intensity of the change in the current longitudinal and / or lateral movement of the vehicle (1) in response to the cut-in situation are set according to the determined level of danger of the cut-in situation.

Images