Method and system for planning the motion of a vehicle

Abstract

A method for planning the motion of a vehicle includes: determining a nominal trajectory for the vehicle based on a desired maneuver to be carried out in a traffic space, on a current state of movement of the vehicle and on a detected state of a surrounding of the vehicle, and determining an abort trajectory branching off from the nominal trajectory and guiding the vehicle to a safe condition regardless of the desired maneuver, wherein the nominal trajectory and the abort trajectory are determined simultaneously using a single optimization process.

Claims

1. A computer-implemented method, the computer-implemented method comprising: determining, by computer-hardware components, a nominal trajectory for a vehicle based on a desired maneuver to be carried out in a traffic space, a current state of movement of the vehicle, and a detected state of a surrounding of the vehicle; and determining an abort trajectory branching off from the nominal trajectory and guiding the vehicle to a safe condition regardless of the desired maneuver, the nominal trajectory and the abort trajectory being determined simultaneously using a single optimization process, the single optimization process determining the nominal trajectory based on the abort trajectory.

2. The computer-implemented method of claim 1, the method further comprising: determining a set of at least two abort trajectories branching off from the nominal trajectory at different points in time along with determining the nominal trajectory in the single optimization process.

3. The computer-implemented method of claim 2, wherein the set of at least two abort trajectories includes at least three abort trajectories branching off from the nominal trajectory at equally spaced points in time.

4. The computer-implemented method of claim 2, wherein using the single optimization process comprises: assigning a cost term to the nominal trajectory and each of the abort trajectories of the set of at least two abort trajectories, the cost term for each of the abort trajectories being multiplied by a respective weight for each of the abort trajectories branching off from the nominal trajectory; and minimizing a cost function, the cost function being a weighted sum of the cost terms.

5. The computer-implemented method of claim 4, wherein assigning the cost term and the weight comprises assigning different weights to at least two of the abort trajectories of the set of at least two abort trajectories.

6. The computer-implemented method of claim 5, wherein a first weight assigned to a first abort trajectory branching off from the nominal trajectory at a first point in time is greater than a second weight assigned to a second abort trajectory branching off from the nominal trajectory at a second point in time which is later than the first point in time.

7. The computer-implemented method of claim 6, wherein the weights assigned to temporally subsequent abort trajectories of the set of at least two abort trajectories are gradually smaller.

8. The computer-implemented method of claim 2, wherein determining the nominal trajectory comprises determining the nominal trajectory as an output trajectory to be output to a vehicle control module unless at least one of the following conditions is satisfied: (i) the nominal trajectory is unsafe; or (ii) a point in time is reached where a last abort trajectory of all subsequent abort trajectories branches off from the nominal trajectory and the nominal trajectory is not confirmed to be safe; and wherein determining the set of at least two abort trajectories comprises determining one of the abort trajectories of the set of at least two abort trajectories as the output trajectory if at least one of the conditions (i) and (ii) is satisfied.

9. The computer-implemented method of claim 8, wherein determining the set of at least two abort trajectories comprises determining a first available abort trajectory as the output trajectory if condition (i) is satisfied, the first available abort trajectory branching off from the nominal trajectory at an earliest point in time after condition (i) is confirmed to be satisfied.

10. The computer-implemented method of claim 9, wherein determining the set of at least two abort trajectories comprises determining a last abort trajectory of all subsequent abort trajectories is determined as the output trajectory if condition (ii) is satisfied.

11. The computer-implemented method of claim 1, wherein the single optimization process comprises a constrained nonlinear process.

12. The computer-implemented method of claim 1, wherein using the single optimization process comprises iteratively executing the single optimization process.

13. A computer system, the computer system comprising a plurality of computer hardware components configured to plan a motion of a vehicle by: determining a nominal trajectory for the vehicle based on a desired maneuver to be carried out in a traffic space, a current state of movement of the vehicle, and a detected state of a surrounding of the vehicle; and determining an abort trajectory branching off from the nominal trajectory and guiding the vehicle to a safe condition regardless of the desired maneuver, the nominal trajectory and the abort trajectory being determined simultaneously using a single optimization process, the single optimization process determining the nominal trajectory based on the abort trajectory.

14. The computer system of claim 13, the computer system further comprising the vehicle.

15. The computer system of claim 14, wherein the plurality of computer hardware components are further configured to determine a set of at least two abort trajectories branching off from the nominal trajectory at different points in time along with the nominal trajectory in the single optimization process.

16. The computer system of claim 15, wherein the set of at least two abort trajectories includes at least three abort trajectories branching off from the nominal trajectory at equally spaced points in time.

17. The computer system of claim 15, wherein the plurality of computer hardware components are further configured to use the single optimization process by: assigning a cost term to the nominal trajectory and each of the abort trajectories of the set of at least two abort trajectories, the cost term for each of the abort trajectories being multiplied by a respective weight for each of the abort trajectories branching off from the nominal trajectory; and minimizing a cost function, the cost function being a weighted sum of the cost terms.

18. The computer system of claim 17, wherein different weights are assigned to at least two of the abort trajectories of the set of at least two abort trajectories.

19. The computer system of claim 18, wherein a first weight assigned to a first abort trajectory branching off from the nominal trajectory at a first point in time is greater than a second weight assigned to a second abort trajectory branching off from the nominal trajectory at a second point in time which is later than the first point in time.

20. A non-transitory computer-readable storage medium comprising computer-executable instructions that, when executed, cause computer hardware components to plan a motion of a vehicle, by: determining a nominal trajectory for the vehicle based on a desired maneuver to be carried out in a traffic space, a current state of movement of the vehicle, and a detected state of a surrounding of the vehicle; and determining an abort trajectory branching off from the nominal trajectory and guiding the vehicle to a safe condition regardless of the desired maneuver, the nominal trajectory and the abort trajectory being determined simultaneously using a single optimization process, the single optimization process determining the nominal trajectory based on the abort trajectory.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) Exemplary embodiments and functions of the present disclosure are described herein in conjunction with the following drawings, showing schematically:

(2) FIG. 1 an illustration of a host vehicle in a traffic space as well as trajectories representing the planning of a motion of the host vehicle according to various embodiments;

(3) FIG. 2 a diagram showing the probability of an abortion of a maneuver as well as an allowed discomfort level within a time horizon; and

(4) FIG. 3 a flow diagram illustrating a method for planning a motion of the host vehicle according to various embodiments.

DETAILED DESCRIPTION

(5) In FIG. 1, a host vehicle 11 is depicted which moves on a road 13 having a continuous lane 14 and an ending lane 15. As shown, the host vehicle 11 moves on the ending lane 15. The host vehicle 11 is provided with an electronic control system including a computer system, not shown. The control system may provide a lane change assistance functionality or may be configured as an autonomous driving system. Environment sensors (not shown) such as a camera, a Radar system and/or a Lidar system for monitoring the environment are installed in the host vehicle 11 and are connected to the control system.

(6) The control system may include a behavior planning module and a motion planning module, for example in the form of integrated circuits. The behavior planning module is configured to send requests regarding desired maneuvers of the host vehicle 11, such as a lane following maneuver, a lane change maneuver or a merge-in maneuver, to the motion planning module. The requests may be driver initiated or may be generated by an autonomous logic.

(7) According to various embodiments, the motion of the host vehicle 11 is planned by determining a nominal trajectory Γ.sub.0 based on the desired maneuver, on the current state of movement of the host vehicle 11 and on the surrounding of the host vehicle 11 as detected by the environment sensors. In FIG. 1, the desired maneuver is a merge-in from the ending lane 15 to the continuous lane 14.

(8) For the generation of the nominal trajectory Γ.sub.0, a time horizon may be defined which corresponds to a range of the perception capability of the environment sensors. The time horizon is divided into equal time increments extending between points in time t.sub.0, t.sub.1, t.sub.2, . . . to provide nodes 19 for the trajectory planning. Each node 19 may include a position in longitudinal direction, a position in lateral direction and a yaw angle of the host vehicle 11 at a point in time t.sub.i.

(9) Apart from the host vehicle 11, other traffic participants such as a passenger car 20 and a truck 22 move on the road 13. As illustrated by the line of sight 25, the passenger car 20 is occluded by the truck 22. Since the velocity of the passenger car 20 might be high enough for a collision with the lane changing host vehicle 11, it can't be guaranteed that the maneuver according to the nominal trajectory Γ.sub.0 is safe. Therefore, a set of abort trajectories (fail-safe trajectories) Γ.sub.a_ti branching off from the nominal trajectory Γ.sub.0 and guiding the host vehicle 11 to a safe condition regardless of the desired maneuver are determined. In FIG. 1, two abort trajectories Γ.sub.a_t1, Γ.sub.a_t2 are shown. However, more than two abort trajectories may be generated for the nominal trajectory Γ.sub.0. As shown, the abort trajectories Γ.sub.a_t1, Γ.sub.a_t2 branch off from the nominal trajectory Γ.sub.0 at subsequent nodes 19, i.e. at different points in time t.sub.i. The starting state of each abort trajectory Γ.sub.a_ti is identical to the state of the nominal trajectory Γ.sub.0 at the starting time of the abort trajectory Γ.sub.a_ti, i.e. Γ.sub.0(t.sub.i)=Γ.sub.a_ti(t.sub.i). According to the example shown, the abort trajectories Γ.sub.a_t1, Γ.sub.a_t2 guide the host vehicle 11 back to the ending lane 15. In other words, the abort trajectories Γ.sub.a_t1, Γ.sub.a_t2 correspond to an abortion of the lane change maneuver.

(10) According to various embodiments, the nominal trajectory Γ.sub.0 and the abort trajectories Γ.sub.a_t1, Γ.sub.a_t1 are determined simultaneously using a single nonlinear optimization process. The nonlinear optimization process includes the step of minimizing a cost function, for example the following cost function ƒ(x):

(11) $f\left(x\right)=f_{\mathrm{\Gamma 0}}\left(x\right)+\underset{i=1}{\overset{N}{.Math.}}{w}_i{f}_{\Gamma a_{-}\mathrm{ti}}\left(x\right)$

(12) wherein x is a vector including motion state parameters such as position and velocity parameters, ƒ(x) is the overall cost function, ƒ.sub.Γ0(x) is the cost function of the nominal trajectory Γ.sub.0, ƒ.sub.Γa_ti(x) is the cost function for the abort trajectory Γ.sub.a_ti branching off from the nominal trajectory Γ.sub.0 at the point in time t.sub.i and w.sub.i is the weight of the abort trajectory Γ.sub.a_ti branching off from the nominal trajectory Γ.sub.0 at the point in time t.sub.i.

(13) The continuity and the smoothness of a motion which starts according to the nominal trajectory Γ.sub.0 and continues with an abort trajectory Γ.sub.a_ti may be enforced by equality constraints defined for each abort trajectory Γ.sub.a_ti. The equality constraints may include:
for i in i=1 . . . Γ.sub.0(t.sub.i)=Γ.sub.at.sub.i(t.sub.i);{dot over (Γ)}.sub.0(t.sub.i)={dot over (Γ)}.sub.at.sub.i(t.sub.i);{umlaut over (Γ)}.sub.0(t.sub.i)={umlaut over (Γ)}.sub.at.sub.l(t.sub.i).

(14) The weights w.sub.i assigned to temporally subsequent abort trajectories Γ.sub.a_ti are gradually decreasing. Therefore, the allowed discomfort level for a late abort trajectory is higher than the allowed discomfort level for an early abort trajectory. An exemplary qualitative level of allowed discomfort over the time horizon as tuned by the weights w.sub.i is shown in the lower panel of FIG. 2. The upper panel of FIG. 2 shows the continuously decreasing probability of an abortion of the maneuver over the time horizon. For rarely occurring late abortions, a higher level of discomfort, for example a jerk, is acceptable.

(15) The nominal trajectory FO is executed, i.e. determined as an output trajectory to be outputted to a control module, until one of the following conditions is fulfilled:

(16) (i) the nominal trajectory turns out to be unsafe, for example due to a detection of the passenger car 20 approaching with high velocity; and

(17) (ii) a point in time t.sub.2 is reached where the last available abort trajectory Γ.sub.a_t2 branches off from the nominal trajectory Γ.sub.0 and the nominal trajectory Γ.sub.0 is neither confirmed to be safe nor to be unsafe.

(18) If condition (i) is fulfilled, the first available abort trajectory, i.e. the abort trajectory planned for lowest t.sub.i>t.sub.c is executed after reaching t.sub.i, wherein t.sub.c is the point in time when the nominal trajectory Γ.sub.0 is confirmed to be unsafe.

(19) If condition (ii) is fulfilled, the last available abort trajectory is executed.

(20) If the situation can be confirmed to be safe before reaching t.sub.2, for example due to the detection of the target lane 14 being empty, the entire nominal trajectory Γ.sub.0 is executed.

(21) FIG. 3 shows a flow diagram illustrating a method for planning a motion of the host vehicle 11 (FIG. 1) according to various embodiments. In step 51, the nominal trajectory Γ.sub.0 and the abort trajectories Γ.sub.a_t1, Γ.sub.a_t2 are determined simultaneously using the single nonlinear optimization process as described above. In step 52, the nominal trajectory Γ.sub.0 is executed. In step 53, it is checked if the nominal trajectory Γ.sub.0 is confirmed to be safe. If it is confirmed to be safe, the process returns to step 52. If the nominal trajectory Γ.sub.0 is not confirmed to be safe, it is checked in step 54 if the nominal trajectory Γ.sub.0 is confirmed to be unsafe. If it is confirmed to be unsafe, the first available abort trajectory Γ.sub.a_ti is executed in step 55. Otherwise, it is checked in step 56 if a point in time is reached or approaching where the last available abort trajectory Γ.sub.a_ti branches off from the nominal trajectory Γ.sub.0. If this point in time is reached or approaching, the last available abort trajectory Γ.sub.a_ti is executed in step 57. Otherwise, the nominal trajectory Γ.sub.0 is executed until the maneuver is finished. The method illustrated in FIG. 3 may be carried out by hardware components of the computer system of the host vehicle 11 (FIG. 1).

(22) In contrast to an independent planning of the nominal trajectory Γ.sub.0 and the abort trajectories Γ.sub.a_ti, wherein the nominal trajectory Γ.sub.0 is planned before the abort trajectories Γ.sub.a_ti, a common planning in a single optimization process enables the nominal trajectory Γ.sub.0 to be shaped in a way that states without any safe abort trajectories are avoided. The planning in a partially unknown environment is thus simplified. Complex maneuvers without complete information regarding the static and dynamic environment thanks to the guaranteed existence of safe abort trajectories may be undertaken. Assigning different comfort cost multipliers and constraints to subsequent abort trajectories planned with a nominal trajectory allows to tune the system in a way that assertive maneuvers are allowed while the most plausible early abort trajectories are kept comfortable for passengers of the host vehicle 11.