Method for steering a vehicle

Abstract

For steering a vehicle around obstacles, proceeding along a path from a starting position to an end position, the path including linear sub-paths which are defined by an increment η and a steering angle δ, a method includes: a) determining a maximum steering angle range and a maximum and a minimum increment range; b) determining the present distance e.sub.P from the end position, the target angle θ.sub.O, and the angle difference e.sub.θ between the present vehicle angle and the target angle; c) performing an optimization method for ascertaining a sub-path by minimizing the value of a cost function l.sub.o assigned to the sub-path; d) determining the new position by adding the ascertained sub-path to the present position; and e) repeating steps (b) through (d) until the end position is reached.

Claims

1. A method for steering a vehicle around at least one obstacle, the vehicle being steered along a path extending from a starting position to an end position, wherein the path includes multiple linear sub-paths which are each defined by an increment η and a steering angle δ, the method comprising: a) determining a maximum steering angle range, a maximum increment range, and a minimum increment range; b) determining a present distance e.sub.P from the end position, a target angle θ.sub.O, and an angle difference e.sub.θ between a present vehicle angle θ and the target angle θ.sub.O; c) carrying out an optimization method for ascertaining a linear sub-path by minimizing the value of a cost function l.sub.O assigned to the linear sub-path, wherein: (i) the cost function includes the present distance e.sub.P from the end position and the present angle difference e.sub.θ to the target angle as optimization variables which are weighted independently of each other, and (ii) the maximum steering angle range, the maximum increment range, the minimum increment range, and a collision test are provided as boundary conditions; d) determining a new position by adding the ascertained linear sub-path to the present position; and e) repeating steps (b) through (d) until the end position is reached f) steering the vehicle along the path in accordance with at least one of the ascertained linear sub-paths.

2. The method as recited in claim 1, wherein the entire path to the end point is initially calculated, and subsequently the vehicle is steered along the path.

3. The method as recited in claim 1, wherein the vehicle is steered along the calculated linear sub-paths during the calculation of the overall path to the end point.

4. The method as recited in claim 1, wherein the method is divided into at least two phases as a function of the position of one corner of the vehicle relative to at least two obstacles, a target angle θ.sub.O=α which is essentially normal to a boundary line between the at least two obstacles being selected in a first phase.

5. The method as recited in claim 4, wherein the at least two phases differ by weighting factors of the optimization variables.

6. The method as recited in claim 5, wherein the method is divided into two phases A and B, and wherein: a) in phase A, the weighting of the present distance e.sub.P from the end position outweighs the weighting of the angle difference e.sub.θ to the target angle to quickly move the vehicle in the direction of the end position; b) in phase B, the weighting of the angle difference e.sub.θ to the target angle outweighs the weighting of the distance e.sub.P from the end position to bring the vehicle into the correct angular position; and c) a switch is carried out between phase A and phase B at a switching point as soon as the position of one corner of the vehicle passes through a boundary line between the at least two obstacles.

7. The method as recited in claim 6, wherein, during the optimization method, points of change in direction are determined at which the direction of the sub-path to be determined is changed, the points of change in direction being defined as points at which one of: a) in phase A, no sub-path is ascertainable whose associated cost function is improved compared to the previously ascertained sub-path; or b) in phase B, no sub-path is ascertainable which is within the allowed increment range.

8. The method as recited in claim 5, wherein, during the optimization method, points of change in direction are determined at which the direction of the sub-path to be determined is changed, the points of change in direction being defined as points at which one of: a) no sub-path is ascertainable whose associated cost function is improved compared to the previously ascertained sub-path; or b) no sub-path is ascertainable whose increment is within the allowed increment range.

9. The method as recited in claim 1, wherein the method is divided into at least two phases as a function of the position of one corner of the vehicle relative to at least two obstacles, a target angle θ.sub.O=α which is essentially parallel to the longitudinal extension of the at least two obstacles being selected in a first phase.

10. The method as recited in claim 1, wherein the method is divided into at least two phases as a function of the position of one corner of the vehicle relative to at least two obstacles, a target angle θ.sub.O=θ.sub.S which is essentially parallel to a boundary line between the at least two obstacles being selected in a second phase.

11. A non-transitory, computer-readable data storage medium storing a computer program having program codes which, when executed on a computer, perform a method for steering a vehicle around at least one obstacle, the vehicle being steered along a path extending from a starting position to an end position, wherein the path includes multiple linear sub-paths which are each defined by an increment η and a steering angle δ, the method comprising: a) determining a maximum steering angle range, a maximum increment range, and a minimum increment range; b) determining a present distance e.sub.P from the end position, a target angle θ.sub.O, and an angle difference e.sub.θ between a present vehicle angle θ and the target angle θ.sub.O; c) carrying out an optimization method for ascertaining a linear sub-path by minimizing the value of a cost function l.sub.O assigned to the linear sub-path, wherein: (i) the cost function includes the present distance e.sub.P from the end position and the present angle difference e.sub.θ to the target angle as optimization variables which are weighted independently of each other, and (ii) the maximum steering angle range, the maximum increment range, the minimum increment range, and a collision test are provided as boundary conditions; d) determining a new position by adding the ascertained linear sub-path to the present position; and e) repeating steps (b) through (d) until the end position is reached f) steering the vehicle along the path in accordance with at least one of the ascertained linear sub-paths.

12. A control system for steering a vehicle around at least one obstacle, the vehicle being steered along a path extending from a starting position to an end position, wherein the path includes multiple linear sub-paths which are each defined by an increment η and a steering angle δ, the control system comprising: a detection system including sensors for ascertaining the starting position, the end position, the steering angle δ, the vehicle angle θ, and the coordinates of the at least one obstacle; and a control unit including a processor configured to perform the following: a) determining a maximum steering angle range, a maximum increment range, and a minimum increment range; b) determining a present distance e.sub.P from the end position, a target angle θ.sub.O, and an angle difference e.sub.θ between a present vehicle angle θ and the target angle θ.sub.O; c) carrying out an optimization method for ascertaining a linear sub-path by minimizing the value of a cost function l.sub.O assigned to the linear sub-path, wherein: (i) the cost function includes the present distance e.sub.P from the end position and the present angle difference e.sub.θ to the target angle as optimization variables which are weighted independently of each other, and (ii) the maximum steering angle range, the maximum increment range, the minimum increment range, and a collision test are provided as boundary conditions; d) determining a new position by adding the ascertained linear sub-path to the present position; and e) repeating steps (b) through (d) until the end position is reached f) steering the vehicle along the path in accordance with at least one of the ascertained linear sub-paths.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 shows a schematic outline of the kinematic model used for the vehicle.

(2) FIGS. 2a through 2b show schematic outlines of an operation of pulling a vehicle out of a parking space carrying out the method according to the present invention.

(3) FIG. 3 shows a schematic representation of different parking scenarios.

(4) FIG. 4 shows a schematic representation of one parking scenario.

(5) FIG. 5 shows a representation of the method according to the present invention using pseudocode.

DETAILED DESCRIPTION OF THE INVENTION

(6) FIG. 1 shows a schematic outline of the kinematic model used for the vehicle. Vehicle 1 (not shown) has a length l.sub.c and a width w.sub.c, for example corresponding to the illustrated table, and is moving in the direction of vector v and initially has vehicle orientation θ with respect to the used coordinate system. The two wheels of one axis are combined here to each form an imaginary wheel 10 in the center of the axis, the front wheel having steering angle δ. The maximum steering angle in this exemplary embodiment is 45°. The distance between the two wheels is the corresponding wheelbase of vehicle L.

(7) FIG. 2a shows a vehicle 1 which is in a starting position 2 and is to be steered into an end position 3 without coming in contact with obstacles 7. A number of sub-paths 4 are determined for this purpose, a static optimization problem being solved for each sub-path.

(8) As a function of the position of a corner 5 of the vehicle relative to an imaginary boundary line 6 between obstacles 7, a distinction is made between phases A and B in which different weightings are used for the particular optimization method to be solved. In the illustrated case, vehicle 1 is in the parked state in phase B and is to be moved out of the parking space with preferably strong turning movements, without the absolute distance from end point 3 being of high importance. For this reason, target angle θ.sub.O=α is set in the optimization method, and the weighting of the distance from end position e.sub.P is drastically reduced compared to the weighting of the angle difference to target angle e.sub.θ=θ−θ.sub.O.

(9) FIG. 2b shows vehicle 1 just prior to reaching end position 3. Corner 5 has traversed boundary line 6 at switching point 8, and a switch was made from phase B to phase A. Correspondingly, target angle θ.sub.O=θ.sub.S was set in the optimization method, and the weighting of the distance from end position e.sub.P was drastically increased compared to the weighting of the angle difference to target angle e.sub.θ.

(10) FIG. 3 schematically shows different scenarios of the method, namely parallel pulling into and out of a parking space or parallel parking (I), pulling into and out of a garage or cross parking (II), and angled pulling into and out of a parking space (III). In each scenario, a possible relevant corner 5 of the vehicle is marked, which may be used to determine the phase change between phase A and phase B. Other corners are also conceivable.

(11) The range of phases A and B is also shown. In all three scenarios, target angle θ.sub.S in phase A points in parallel to boundary line 6. In contrast, target angle α in phase B is normal to boundary line 6 in scenarios I and II, but in parallel to the longitudinal extension of obstacles 7 in scenario III.

(12) FIG. 4 shows a schematic representation of a parking scenario in which the driving direction is changed. Proceeding from starting point 2, vehicle 1 initially moves out of a parking space and thus moves out of phase B into phase A. Thereafter, however, a situation occurs in which the vehicle moves further away from end point 3 by every forward movement.

(13) Consequently no sub-path whose cost function is lower than the previously calculated cost function is ascertainable in the optimization method in phase A.

(14) This situation is recognized at point of change in direction 9 at which the vehicle changes the driving direction and backs up. The vehicle consequently approaches end point 3 again. Of course it may also be necessary to change the driving direction multiple times during an operation of pulling into or out of a parking space to obtain results of the optimization method which are always valid and continuously reduced.

(15) FIG. 5 shows a detailed representation of the method according to the present invention using pseudocode. In this exemplary embodiment, a method for pulling out of a parking space is carried out, and the starting position of vehicle q.sub.0 is initialized as parking position q.sub.P. Thereafter, target angle θ.sub.O is set to value α. The value of a depends on the particular parking situation. Both in the case of parallel parking and in the case of cross parking, the vehicle is to be moved normally to boundary line 6 if possible for the operation of pulling out of the parking space.

(16) Thereafter, weight term r.sub.θ and weight matrix R are set. Depending on the phase, the initial values r.sub.θ, θ.sub.O and R are set as follows, for example, it also being possible, of course, for other weightings to be provided:

(17) $\hspace{1em}\begin{array}{ccc}& \mathrm{Phase}A& \mathrm{Phase}B\\ r_{\theta }& 4& 1\\ {\theta }_O& {\theta }_S& \alpha \\ R& \left(\begin{array}{cc}30& 0\\ 0& 1\end{array}\right)& \left(\begin{array}{cc}0.1& 0\\ 0& 0\end{array}\right)\end{array}$

(18) In the present exemplary embodiment, an operation for pulling out of a parking space is determined; the method is thus initially in phase B. Direction D of the vehicle is initialized with 1 (forward) and starting solution u=[u.sub.l0, η.sup.0].sup.T for steering input u.sub.l=1/L tan(δ) and increment η. Vectors u.sub.min and u.sub.max limit the steering input and the increment.

(19) Thereafter, the optimization method is run through in a loop until the position of the vehicle is in an epsilon surroundings around the end point and thus the end position has been reached.

(20) It is checked in every passage of the loop whether the position of corner 5 passes through the imaginary boundary line of the obstacles (x.sub.CFI,i≧b), and in this case the weighting terms of phase A from above the table are used.

(21) Thereafter, the actual optimization method is started, which is defined as follows:

(22) $\underset{u_i}{\min }{l}_{O_i}\left(q_{i+1}\right)$$s.t.q_{i+1}=f\left(q_i,u_i,D\right)$$h_P\left(q_{i+1}\right)\le 0$$u_{\min }\le u_i\le u_{\max },$
the cost function to be minimized reading:
l.sub.O.sub.i(q.sub.i+1)=Tθe.sub.0.sub.i+1.sup.2+e.sub.P.sub.i+1.sup.TRe.sub.P.sub.i+1.

(23) In this method, q.sub.i denotes the present position of the vehicle in relation to step i, u.sub.i denotes the increment vector, and D denotes the direction. The cost function is minimized with respect to variables e.sub.θl and e.sub.Pl. The term e.sub.Pl=[x.sub.i−x.sub.s, y.sub.i−y.sub.s].sup.T denotes the distance between the vehicle and starting position [x.sub.s, y.sub.s] at step i (since the entire path is calculated backward from the end position to the starting position). The term e.sub.θl=θ.sub.i−θ.sub.O denotes the difference between the present vehicle angle and target angle θ.sub.O. Parameters r.sub.θ and R are weighting parameters.

(24) The function h.sub.P(q.sub.i) denotes a collision detection which detects a contact of the given dimensions of the vehicle with the obstacles on the geometric path. Finally, u.sub.min and u.sub.max denote the minimum and maximum values of the optimization variables made up of the steering input and the increment.

(25) In every iteration step, the position and the vehicle angle are improved with respect to the cost function, the cost function fulfilling the purpose of allowing the starting point to be reached rapidly by selecting defined weighting parameters r.sub.θ and R in phases A and B.

(26) New position q.sub.i+1 is represented by a movement equation f(q.sub.i, u.sub.i, D), which reads

(27) $\begin{array}{c}{q}_{i+1}=\left(\begin{array}{c}{x}_{i+1}\\ y_{i+1}\\ {\theta }_{i+1}\end{array}\right)\\ =\left(\begin{array}{c}{x}_i+D{\eta }_i\cos \left({\theta }_i+D\frac{{\eta }_i{u}_{l_i}}{2}\right)\\ y_i+D{\eta }_i\cos \left({\theta }_i+D\frac{{\eta }_i{u}_{l_i}}{2}\right)\\ {\theta }_i+D{\eta }_i{u}_{l_i}\end{array}\right)\\ =f\left(q_i,u_i,D\right)\end{array},$

(28) The optimization problem of cost function I.sub.Oi(q.sub.i) is always formulated only for one incremental step after another and is repeated until the end point is reached. After every optimization, it is checked whether either the value of the cost function is greater than in the preceding optimization (l.sub.Oi*>k.sub.i.Math.l.sub.Ol−l*) or a collision takes place (h.sub.P(q.sub.i+l*)>0) and thus no sub-path was found. Here, k.sub.l denotes a weighting factor to also allow sufficiently small degradations. If this is the case, the direction of the sub-path is reversed by equating D to −D.

(29) Otherwise, new position q.sub.i+1 is ascertained with the aid of movement equation f (q.sub.i, u.sub.i, D), and the method is continued with the new position. The method is ended when the new position is in an epsilon surroundings around the end point.

(30) The method according to the present invention may be part of control electronics of a vehicle or be present in the form of a computer program on a data carrier. The accordingly steered vehicle is equipped with sensors for detecting the surroundings, and preferably also the host vehicle's driving movements. A GPS sensor and multiple ultrasonic sensors or cameras may be provided, for example, to detect the host vehicle's position and the positions of the obstacles. Based thereon, it is possible to determine the required weighting terms for the method.