Automatic forward parking in perpendicular parking spaces

[0001] This patent application is a U.S. National Phase of International Patent Application No. PCT/EP2010/007091, filed 23 Nov. 2010, which claims priority to German Patent Application No. 10 2009 060 169.4, filed 23 Dec. 2009, the disclosures of which are incorporated herein by reference in their entirety.

[0002] Embodiments of the present disclosure relate to a method for the automatic forward parking of a motor vehicle in a perpendicular parking space and to a corresponding driver assistance system.

[0003] In current parking assist systems or driver assistance systems for automatic parking, in which case parking both in the longitudinal parking spaces and in perpendicular parking spaces is carried out here, a parking space is measured using a suitable sensor system as the parking space is passed, and reverse parking is then carried out. With longitudinal parking spaces, this is the only possible way of parking in a longitudinal parking space since the rear axle generally cannot be steered. Longitudinal parking spaces in which forward parking is possible are so large, however, that it would be necessary to refer to a lane change, rather than parking.

[0004] Such a reverse parking operation into a parking space is described, for example, in DE 10 2009 006 336 A1. There, an automatic parking operation of a motor vehicle is monitored, in which case, before the actual parking operation, a parking space is measured using a camera-based method as the motor vehicle passes the parking space, and obstacles in the environment of the parking space are also determined. During the automatic parking operation, the camera-based method is still used to measure the environment of the motor vehicle, the obstacles newly detected in the measurement during the parking operation being compared with previously detected obstacles and a corresponding measure being carried out if there is a discrepancy between newly detected obstacles and known obstacles.

[0005] A method for automatically parking a motor vehicle in a parking space or for moving a motor vehicle out of a parking space can also be gathered from DE 10 2009 006 331 A1. In this case, the parking space is determined by an environment detection system of the motor vehicle and a target trajectory and a position end point relative to the parking space on the target trajectory are defined. A suitable steering angle is then calculated on the basis of the current vehicle position and this steering angle is used for parking or leaving a parking space.

[0006] The common feature of all of these known methods is that, after the parking space has been measured, a target trajectory is defined and is iteratively readjusted if necessary, along which trajectory the motor vehicle reverses into the parking space in one or more maneuvers.

[0007] The disadvantage of the reverse-parking strategies is that, on the one hand, the subsequent traffic is hindered during reverse-parking into a perpendicular parking space and there is also the risk of a subsequent vehicle taking the free parking space by means of forward parking and of forward parking usually being carried out both in parking garages and in garages.

[0008] Therefore, the disclosed embodiments are based on specifying a method for the forward parking of a motor vehicle in a perpendicular parking space, including a garage.

[0009] This is achieved by means of a method without previous measurement of the parking space having the features of claim 1, a method with previous measurement of the perpendicular parking space according to the features of claim 10 and a driver assistance apparatus for carrying out the methods having the features of claim 12.

[0015] In a first disclosed embodiment, the method for the forward parking of a motor vehicle in a perpendicular parking space, the motor vehicle having environment sensors for determining environment data and obstacles in the environment of the motor vehicle, includes:

• • [0016] prealigning the motor vehicle in front of the perpendicular parking space in such a manner that a setpoint steering angle is between a maximum steering angle δ_max and a minimum steering angle δ_min,
  • [0017] iteratively searching the permissible steering angle range between the maximum steering angle δ_max and the minimum steering angle δ_min for a current steering angle δ_curr,i during a forward maneuvering movement of the motor vehicle into the perpendicular parking space, the current steering angle δ_{curr, i} giving rise to a maximum free path length s_i, i=0, . . . n, into the perpendicular parking space without the vehicle hitting obstacles, and
  • [0018] terminating the forward maneuvering movement if the end of the parking operation has been reached or a reverse maneuvering movement must be carried out on account of an obstacle.

[0019] In this case, the maximum steering angle δ_max and minimum steering angle δ_min may be in the range ±17°.

[0020] Obstacles which have been found are optionally recorded, that is to say stored, in a map of the environment by the environment sensors. Obstacles which disappear again are likewise removed again from the map of the environment.

[0021] The first step i=0 of the iterative search may be carried out by scanning a predefined curvature range κveh;i=0 with a predefined curvature iteration size Δκ in order to determine an ideal curvature estimated value, and the forward maneuvering movement of the motor vehicle is carried out along the ideal curvature estimated value of the first step. In this case, the curvature estimated value of the first step corresponds to the maximum free path length si=0. Furthermore, the curvature estimated value determines the steering angle of the vehicle.

[0022] The predefined curvature range of the first step comprises the range of −0.15 to 0.15, the curvature iteration size Δκ of the first iteration step being 0.0005. In this case, the predefined curvature range of −0.15 to +0.15 corresponds to a minimum steering angle and a maximum steering angle of approximately ±17° for a wheel base of approximately 2 m.

[0023] For the second step and the subsequent steps i=1, 2, . . . , n of the iteration, the curvature estimated value of the previous step is used as the initial value for the current step, scanning being carried out around the initial value of the previous step with a current curvature iteration size to determine the current curvature estimated value of the i-th step, the current curvature iteration size being a function of the maximum path length si-1 of the previous step. Furthermore, the curvature range to be searched in the current step to determine the current curvature estimated value is a function of the curvature iteration size of the current step, and the forward maneuvering movement of the motor vehicle is carried out along the current curvature estimated value of the second step and the subsequent steps.

[0024] The curvature iteration size of the second step and the subsequent steps is optionally determined by means of the following formula:

[0025] $\begin{array}{cc}\Delta \kappa =\frac{\Delta {\kappa }_0}{{\left(\min \left(s_{i-1},{\sigma }_0\right)/{\sigma }_0\right)}^2}& \left(1\right)\end{array}$

[0026] In this case, s_i-1 is the maximum distance determined in the previous step and σ₀ is an experimentally determined constant. In the present case, σ₀=3 m has proved to be a reasonable value.

[0027] To determine the current curvature estimated value, the curvatures, that is to say the following curvature range, are optionally searched:
κ_search,j=κ₀+(j−3)Δκ  (2)

[0028] In this case, optionally j=0, 1, . . . m, where m is a natural number >0. Optionally, m=6, in other words the search range comprises seven values. Other values for m are naturally possible, as a result of which the search range becomes larger or smaller. Furthermore, K₀ is defined as the curvature of the previous step, that is to say
κ₀=κ_veh,i-1  (3)
Furthermore, it is possible to determine the vehicle alignment relative to the parking space by considering the detected lateral obstacles which are within a predefined distance value, a left-hand regression line and a right-hand regression line respectively being placed through the obstacle points defined by distances between the lateral, that is to say left-hand and right-hand, obstacles. The position and alignment of the vehicle relative to the perpendicular parking space can then be determined from the two regression lines by means of averaging and by considering the enclosed angle. In this case, the position and alignment during the forward maneuvering movement can likewise be taken into account. For the reverse maneuvering movement, the detection of the alignment of the vehicle in the parking space and, therefrom, the average distances from the vehicle to the left and the right is used to look for a favorable starting position for the subsequent forward maneuvering movement.

[0029] A second disclosed embodiment of the method for the forward parking of a motor vehicle in a perpendicular parking space involves the motor vehicle having environment sensors for determining environment data and obstacles in the environment of the motor vehicle, a perpendicular parking space being measured by the environment sensors as the motor vehicle passes the perpendicular parking space. A parking trajectory for the forward parking of the motor vehicle relative to the current location of the motor vehicle is then calculated, the motor vehicle being aligned by a reverse movement in such a manner that it can park in the perpendicular parking space with a subsequent forward movement. For the iterative forward parking of the motor vehicle in the perpendicular parking space, the vehicle can move along the calculated trajectory, the trajectory being able to be corrected again and again by means of current environment data. It is also possible for the forward movement of the motor vehicle to be effected by means of the above-described iterative method of the first embodiment.

[0030] As already mentioned, during the automatic parking operation, the environment sensors can still measure the environment of the vehicle, and the parking trajectory can thus be adapted to the new environment data.

[0031] A driver assistance system for the automatic parking of a motor vehicle and for carrying out the methods described above comprises environment sensors for determining environment data relating to the motor vehicle, a calculation unit for continuously calculating a parking trajectory from the environment data, and a controller which moves the motor vehicle. In this case, the controller comprises actuators for accelerating and decelerating the motor vehicle, actuators for a braking intervention and actuators for a steering intervention.

[0032] FIG. 1 shows a vehicle 1 which has a diagrammatically illustrated front axle 2 and rear axle 3 and is in front of a parking space 4. In this case, a steering angle range 5 min and δmax is diagrammatically illustrated in FIG. 1. The steering angle range between δmin and δmax is searched for steering angles which make it possible to travel as far as possible without hitting obstacles. Two possible curvatures, that is to say steering angles, 5, 6 are illustrated in FIG. 1, in which case it is clear that it is possible to travel further into the perpendicular parking space 4 with the curve 5 than with the curve 6. The route which provides the greatest distance without hitting an obstacle, that is to say the curve 5 in the present case, predefines the steering angle, that is to say the curvature, which is then taken in the first step. As a note, it is remarked that travel straight ahead is carried out with the proposed iterative method in the absence of obstacles. Therefore, the method can be used not only in garages but also in confined driving situations.

[0033] At the start of the keying-in algorithm, the entire possible curvature range κi=0=−0.15 . . . 0.15 is scanned in curvature iteration sizes of Δκ=0.0005 to obtain a lower initial value for κi=0. This initial value and also all subsequent values always use the curvature determined last as a good estimation of the current iteration step to restrict the search space for the subsequent steps. This is because, in the subsequent steps, there is a restriction for reasons of the computation power of the search space and a search is carried out for new optimum curvatures only in the area surrounding the curvature determined last. to escape from local minima which are produced, in particular, when the vehicle approaches an obstacle, the curvature iteration size of each step is coupled to the free path length si-1 from the last step, that is to say

[0034] $\begin{array}{cc}\Delta \kappa =\frac{\Delta {\kappa }_0}{{\left(\min \left(s_{i-1},{\sigma }_0\right)/{\sigma }_0\right)}^2}& \left(1\right)\end{array}$
where σ₀=3 m is an experimentally determined constant and s_i-1 is the maximum distance determined from the last step. The local minima, where a current minimum may also be a global minimum, can be recognized from the fact that the maximum distance becomes smaller and smaller. Therefore, the search range is then increased to possibly escape from a local minimum.

[0035] When searching for a new ideal curvature, the curvatures
κ_search,j=κ₀+(j−3)Δκ  (2)
around the curvature from the last step are checked, where j can assume the values j=0, . . . , 6 and K₀ is the curvature found in the preceding step, that is to say K₀=K_veh,i-1.

[0036] With each of these path curvatures, an area on the roadway which is restricted by the vehicle boundaries is defined. Therefore, in the case of a left turn, the left rear corner and the front right corner form the area boundary and, in the case of a right turn, the front left corner and the rear right corner form the area boundary, as is also illustrated by the curves 5, 6 in FIG. 1.

[0037] There are now two possibilities of why a forward movement must be stopped. Either the vehicle has reached the end of the parking operation or the vehicle must carry out a reverse maneuvering movement to circumvent an obstacle.

[0038] FIG. 2a and FIG. 2b thus show two possible single-maneuver parking scenarios for different starting positions of the vehicle 1 in front of the parking space 4. In other words, the driver approaches a favorable starting position for forward parking, that is to say he positions himself favorably in front of a garage, for example, and gives the driver assistance system the signal to carry out the parking operation using the iterative procedure.

[0039] However, during forward parking, the situation may occur in which the steering lock would mean a long collision-free route for the front of the vehicle but the side of the vehicle would collide with an obstacle, for example one of the corners of a garage. If this is the case, the vehicle must first reverse to achieve a favorable starting position.

[0040] This is illustrated in FIG. 3 in which the vehicle 1 would hit the left-hand edge (viewed in the direction of travel) of the perpendicular parking space in the fifth iteration step. Consequently, the vehicle carries out a reverse movement to avoid a collision with an obstacle. In other words, starting from an unfavorable starting position of the vehicle 1 in front of the perpendicular parking space 4, the vehicle carries out a first forward movement 8 until the risk of a collision between a side of the vehicle 1 and a corner 7 of the perpendicular parking space 4 is imminent. To avoid the collision, the vehicle carries out a first reverse movement 9 to maneuver the vehicle 1 into a more favorable new starting position. For this purpose, the last position of the vehicle 1 in the parking space 4 along the first forward movement and its alignment can be taken into account to arrive at a more favorable new starting position. In other words, the longitudinal axis of the vehicle 1 must be brought closer to the longitudinal axis 11 of the perpendicular parking space 4. The vehicle 1 then carries out a new second forward movement 10 into the parking space 4.

[0041] The end of the forward parking operation is reached when either the end of the parking space is reached, in which case a parking space detection defines a destination, or when the driver forwards a corresponding signal to the vehicle. The realization of whether onward travel is blocked with an obstacle can be predefined by means of a minimum distance.

[0042] FIG. 4 finally shows a parking operation after previous measurement of the parking space. In this case, the vehicle 1 is driven past the parking space 4 and has measured the parking space 4 using suitable environment sensors. It was also signaled to the driver that the parking space is large enough. Starting from a starting position A, the driver assistance unit calculates a parking trajectory, a first reverse movement 12 changing the vehicle from the position A to an intermediate position B which is suitable for a forward parking operation. The vehicle 1 is changed to the position C in the parking space 4 with a subsequent forward movement 13. Conventional trajectory planning systems can be used to plan the trajectory of the forward movement, the iterative forward parking method described above likewise being able to be used for the forward trajectory 13.

• [0043] 1 Vehicle
• [0044] 2 Front axle (steerable)
• [0045] 3 Rear axle
• [0046] 4 Parking space
• [0047] 5 Curve
• [0048] 6 Curve
• [0049] 7 Corner of the perpendicular parking space
• [0050] 8 First forward movement
• [0051] 9 First reverse movement
• [0052] 10 Second forward movement
• [0053] 11 Longitudinal axis of the parking space
• [0054] 12 Reverse movement
• [0055] 13 Forward movement
• [0056] A Starting position
• [0057] B Intermediate position
• [0058] C End position