Motor vehicle and method for motion planning of an autonomous motor vehicle

Abstract

Method for motion planning of an autonomous motor vehicle (10), wherein the method comprises the following steps: - Determining (S1) an initial state of the motor vehicle (10) and an environment of the motor vehicle (10) by a sensor device (12) of the motor vehicle (10); - Determining (S2) a target position and a target state of the motor vehicle (10) in the environment by a computing device (14) of the motor vehicle (10); - Determining (S3) configuration samples (20, 20', 20'') in a configuration space (18) by the computing device (14), wherein the configuration samples (20, 20', 20'') are generated depending on a motion model of the motor vehicle (10) and thus impermissible movements of the motor vehicle (10) are excluded, wherein the configuration space (18) represents the environment of the motor vehicle (10) and the configuration samples (20, 20',20'') possible positions of the motor vehicle (10) with a respective corresponding state of the motor vehicle (10), wherein the configuration samples (20, 20', 20'') are determined using a sampling-based algorithm in predefined sampling steps, wherein in each sampling step a plurality of configuration samples (20, 20', 20'') are determined in parallel using at least one graphics processing unit (15) of the computing device (14), wherein the configuration samples (20, 20', 20'') generated in each sampling step are based on the state and position of the configuration samples (20, 20', 20'') of the preceding sampling step, wherein respective sequences of configuration samples (20, 20', 20'') possible paths (22,22') of the motor vehicle (10) through the configuration space (18) to reach the target position and target state;- wherein the configuration space (18) is rasterized into several grid cells (26) (S4) and a grid cell object distance (28) is determined from a center of a respective grid cell (26) to one of the nearest objects (24) in the environment of the respective grid cell (26), wherein a motor vehicle shape of the motor vehicle (10) in the configuration space (18) is modeled as a motor vehicle circle model comprising a circle (30) or a composition of several circles, wherein for a respective grid cell (26) a distance of the motor vehicle (10) to the nearest object (24) in the environment is calculated by a difference of the grid cell object distance (28) and a radius of a circle (30) of the motor vehicle circle model,- evaluating (S5) the possible paths (22, 22') and / or newly generated Configuration sample values ​​(20, 20',20'') after each sampling step by the computing device (14) depending on a path criterion, wherein the path criterion checks at least the calculated distance of the motor vehicle (10) to the objects (24) in the environment for the presence of at least a minimum distance, excluding paths and / or configuration sampling values ​​for which the minimum distance is not present;- providing (S6) at least one possible path (22') as a motion trajectory of the motor vehicle to a control device (16) of the motor vehicle (10) for controlling the autonomous motor vehicle (10) to the target position in the target state;- wherein objects (24) in the environment are classified as low and high objects (24), wherein for high objects (24) the presence of the minimum distance is checked, wherein for low objects (24) the configuration space (18) is subdivided into a height profile grid with slope information, wherein for configuration sampling values ​​(20, 20', 20''),which are located near a low object (24), a wheel position of the motor vehicle is determined from the state of the respective configuration scan values ​​(20, 20', 20'') in relation to the low object (24), wherein, depending on the wheel position, a lateral distance of a respective wheel (30) to the low object (24) is determined, whereby the possible path (22) and / or the configuration scan value for which the lateral distance is below a wheel distance threshold value is excluded.

Description

The invention relates to a method for motion planning of an autonomous motor vehicle. Furthermore, the invention relates to a motor vehicle that is configured to carry out the method. Several strategies for motion planning, or the planning of motion trajectories, are known. Sampling-based algorithms, in particular, have proven advantageous. However, the results of these algorithms are highly dependent on the number of samples, and a large number of sampling points often complicates real-time applications, rendering these algorithms unsuitable for many scenarios. Especially in confined environments containing one or more objects, covering all possible motion paths is often difficult, necessitating strategies to identify potential paths through such environments. From DE 10 2021 120 632 B3 a method for carrying out an autonomous parking procedure of a vehicle is known. From DE 10 2013 220 931 A1, a method and a device for assisting a motor vehicle in driving onto a curb are known. The method comprises, as steps, detecting a curb located in the vicinity of the motor vehicle and modifying at least one operating parameter of the motor vehicle such that, during the driving onto the curb of at least one tire of the motor vehicle, the vibration of the motor vehicle associated with this driving onto the curb is reduced compared to an analogous driving process without this modification. From DE 10 2016 120 660 A1 a method for determining a trajectory for a vehicle in a map is known.The procedure comprises the following steps: generating a map with a network of initial grid cells of a predetermined cell size, transferring data from a non-drivable area to the map, assigning the non-drivable area to at least one initial grid cell, grouping a plurality of initial grid cells into a superior grid cell, and determining the trajectory based on the initial grid cells and the superior grid cells using an A* method, wherein the grouping of a plurality of initial grid cells into a superior grid cell is performed depending on a distance from the non-drivable area, and the determination of the trajectory based on the initial grid cells and the superior grid cells using an A* method includes determining the trajectory of the largest grid cells. From DE 10 2011 080 932 A1, a method for supporting a vehicle driver in performing driving maneuvers is known. The method comprises steps for providing environmental data, evaluating the environmental data and identifying ground-level objects that can be driven over, and generating a signal that initiates the automatic execution of driving maneuvers, and / or issuing control instructions to the driver in response to identified ground-level objects. The object of the invention is to improve motion planning for an autonomous motor vehicle. This problem is solved by the independent patent claims. Advantageous embodiments of the invention are disclosed in the dependent patent claims, in the following description, and in the figures. The invention is based on the idea that a sample-based algorithm is used for motion planning, wherein a motion model of the motor vehicle and a model of the shape of the motor vehicle are provided, wherein the models limit the generation of sample values ​​only to movements and distances to objects that are possible starting from an orientation of the motor vehicle. One aspect of the invention relates to a method for motion planning of an autonomous motor vehicle, in particular for a parking maneuver. The method comprises the following steps: determining an initial state of the motor vehicle and its environment by a sensor device of the motor vehicle; determining a target position and a target state of the motor vehicle in its environment by a computing device of the motor vehicle; and determining configuration samples in a configuration space by the computing device, wherein the configuration samples are generated depending on a motion model of the motor vehicle and thus impermissible movements of the motor vehicle are excluded, wherein the configuration space represents the environment of the motor vehicle and the configuration samples comprise possible positions of the motor vehicle with a respective corresponding state of the motor vehicle.The configuration samples are determined using a sampling-based algorithm in predefined sampling steps, wherein in each sampling step a plurality of configuration samples are determined in parallel using at least one graphics processing unit of the computing device, wherein the configuration samples generated in each sampling step are based on the state and position of the configuration samples of the preceding sampling step, wherein respective sequences of configuration samples define possible paths of the motor vehicle through the configuration space to reach the target position and target state.Furthermore, the configuration space is rasterized into several grid cells, and a grid cell object distance from the center of a respective grid cell to one of the nearest objects in the environment is determined, wherein a motor vehicle shape of the motor vehicle in the configuration space is modeled as a motor vehicle circle model comprising a circle or a composition of several circles, wherein for a respective grid cell, a distance of the motor vehicle to the nearest object in the environment is calculated by means of a difference between the grid cell object distance and a radius of a circle of the motor vehicle circle model.Furthermore, after each sampling step, the computing unit evaluates the possible paths and / or newly generated configuration samples based on a path criterion. This criterion checks whether the calculated distance of the vehicle to surrounding objects meets at least a minimum distance requirement, and excludes paths and / or configuration samples for which this minimum distance is not met. Finally, at least one possible path is provided as the vehicle's trajectory to a control unit for steering the autonomous vehicle to the target position and into the desired state. In other words, a vehicle's sensor system first acquires sensor data that includes information about the vehicle's surroundings and its initial state. The vehicle's surroundings can include, for example, static objects, particularly road boundaries, and / or other vehicles. The vehicle's initial state can be determined, for example, by its orientation within its surroundings, steering angle, wheel position, and / or speed. The sensor system can include, in particular, steering angle sensors, speed sensors, GPS sensors, radar sensors, laser sensors (especially LiDAR sensors), ultrasonic sensors, and / or camera sensors. From the acquired sensor data, a computing unit in the vehicle, which could be, for example, the vehicle's computer, can determine a target position and a target state of the vehicle in its environment. The target position can be a location to which the vehicle is to travel, such as a parking position, and the target state can specify the state of vehicle parameters that should be achieved at the target position. For example, the target state can specify the vehicle's orientation in space, a speed, a wheel position, and / or other vehicle parameters that the vehicle should assume at the target position. The computing device can further be configured to generate a configuration space that virtually represents the vehicle's environment and allows for the determination of the vehicle's possible positions relative to the target position, along with their corresponding states. These positions and their associated states are generated as configuration samples, or samples that encompass a specific configuration at each position. In other words, the configuration samples can include the respective positions in the configuration space and the corresponding configuration or state of the vehicle. Specifically, the state can include the respective steering angles required at the corresponding positions and the vehicle's orientation in space. To analyze the configuration space, a sampling-based algorithm can be used. This algorithm does not generate complete trajectories through the space, but rather individual points of movement within the space, based in particular on the preceding sampling points, i.e., the previous states and positions of the vehicle. This generation of configuration samples is preferably performed using a graphics processing unit, especially one or more graphics processing units (GPUs), which enables massively parallel generation of the configuration samples. Thus, a multitude of possible movements through the configuration space can be determined for diverse situations. Preferably, multiple sampling steps can be performed, with the configuration samples in each sampling step being based on the previously determined states and positions of the preceding sampling step.For example, such a sampling-based method with a configuration space and configuration sampling values ​​is known from WO 2022 / 268323 A1. To limit the computational effort of the sampling algorithm and provide improved motion planning, configuration samples can be generated using a motion model of the vehicle, thus limiting the number of configuration samples in the configuration space. Specifically, only permissible vehicle movements can be defined and allowed within the motion model, preventing the generation or consideration of configuration samples that contradict the motion model. In particular, only configuration samples achievable through steering movements of the vehicle can be generated, thereby excluding, for example, lateral movements.Starting from a given configuration sample value, the motion model can use the corresponding state and position as input and thus determine the configuration samples based on it in the subsequent sampling step. A given sequence of configuration samples can form a possible geometric path of the vehicle through the configuration space. Here, geometric path refers to the spatial distance to the target position, which does not yet include a speed profile. However, the individual configuration samples can contain speed information, which is needed, for example, to move from one configuration sample to the next, such as when steering is required while the vehicle is stationary. To maintain sufficient distance to objects in the surroundings, especially static objects, the configuration space can be divided into several grid cells, and for each grid cell, a grid cell object distance is determined from the center of the respective grid cell to the nearest object in the surroundings. This means that for each grid cell, a value is available that provides the distance to the nearest object. Furthermore, the shape of the vehicle can be assumed to be a schematic model, thus enabling a quick distance check for each grid cell. Preferably, a simplified model can be provided for this purpose, allowing for a rapid calculation of the vehicle's distance for each grid cell; in particular, the vehicle shape can be modeled as a vehicle circle model. With the vehicle circle model, for example,The shape of the entire vehicle can be simplified and assumed to be a circle with a predetermined radius. Preferably, the vehicle shape can also be represented as a composite of several circles, which may, for example, have different radii. In this case, the accuracy can be increased by adding more circles to the circular model of the vehicle. Using the grid cell object spacing and the vehicle circle model, the distance of the vehicle to the objects can be calculated once for each grid cell. Consequently, for each configuration sample value reaching that cell, the distance to the objects can be estimated, resulting in a time saving when sampling the entire configuration space. Specifically, for each grid cell, the distance of the vehicle to the respective object can be calculated by subtracting the radius of a circle in the vehicle circle model from the grid cell object spacing. That is, the grid cell object spacing can be determined by `min(distance_object)`, where `min(distance_object)` is the minimum distance to the respective objects in the vicinity, and the distance of the vehicle to the objects can be calculated for each grid cell by `min(distance_object)`. After the possible paths and / or configuration samples have been generated in each sampling step, the paths can be evaluated according to a path criterion. The path criterion can, for example, be a condition and / or a requirement that the respective path and / or configuration sample should or must fulfill. The path criterion can, at a minimum, check the calculated distance of the vehicle to objects in the environment for the presence of at least a minimum distance, whereby paths and / or configuration samples for which the minimum distance is not met can be excluded. This means that the corresponding configuration samples can be discarded and not pursued further, so that subsequent configuration samples based on these configuration samples are not generated in the first place, resulting in a time saving in the calculation. Preferably, the path criterion can query additional conditions; for example, the paths can be further evaluated depending on a distance traveled, a number of required steering movements, or changes of direction, in order to create a cost function and determine the best paths. Particularly preferably, the configuration sampling values ​​can be determined and the resulting paths subsequently evaluated in each sampling step, whereby the best-rated paths can be pursued iteratively. This can be carried out until the paths lead to the target position. Once the possible paths to the target position have been determined, one path can be selected from all possible paths based on their evaluation, preferably the highest-rated path. This can be determined in particular using a loss function (or cost function). Once the path has been selected, a motion trajectory can be created using a velocity profile. In other words, the motion trajectory represents the geometric path through space, including the associated velocity along the path. This motion trajectory can then be provided to a vehicle control system designed to autonomously steer the vehicle into the parking position. The invention offers the advantage of improved representation of the static environment, enabling trajectory calculations that maintain a minimum distance to static objects while simultaneously utilizing all available space, which can be particularly important in confined spaces. This allows for runtime-optimized and improved collision detection, enabling fast and real-time motion planning. Furthermore, according to the invention, objects in the environment are classified as low or high. For high objects, the presence of the minimum distance is checked. For low objects, the configuration space is divided into a height profile grid with inclination information. For configuration samples located near a low object, the wheel position of the vehicle is determined from the state of the respective configuration samples relative to the low object. Depending on the wheel position, a lateral distance of the respective wheel to the low object is determined. The possible path and / or configuration sample for which the lateral distance is below a wheel distance threshold is excluded. In other words, the objects in the vicinity of the vehicle can be divided into high and low objects.For this purpose, the determined height of the respective object can be used, whereby objects under twenty centimeters can be classified as low objects, and those over twenty centimeters as high objects. If a high object is present in the vicinity, the minimum distance can be checked using the path criterion described above, whereby the configuration sampling value can be excluded if the minimum distance is too small. However, if a low object is present, it can be checked whether driving over this low object is possible and whether this object poses a danger to the vehicle, especially to the vehicle's rims. Low objects can include, for example, curbs, wheel stops, and speed bumps.To check whether it is possible to drive over a low object and / or whether the low object could damage a wheel rim, the configuration space can be divided into a height profile grid. This also allows for the provision of information such as the inclination of the low objects. A height profile grid means that the configuration space can be divided into multiple units, for example, squares, where each unit of the grid can have a value indicating the height of the object at that position. The inclination information can be provided from values ​​or a profile of adjacent grid cells. Using the height profile grid, it is then possible to check whether there is a danger to a wheel, especially a rim of the vehicle, by first reading the wheel position or orientation from the configuration sample values.Subsequently, the lateral distance of each wheel to the low object can be determined. Lateral distance refers to the distance measured laterally to the object along the normal of the wheel's surface. If it is determined that the lateral distance is below a wheel distance threshold, and thus, for example, the rim could graze the low object, the configuration sample value and / or the associated path can be excluded from motion planning. This has the advantage of allowing for further improvement of motion planning. The invention also includes further developments that result in additional advantages. A further training provision stipulates that paths whose distance to objects is below a predefined safety distance value are rated lower by the path criterion than paths above the safety distance value. The safety distance value can be greater than the minimum distance required by the path criterion. However, the safety distance value can provide a safety margin that should be taken into account, for example, due to inaccuracies in motion planning. For instance, a minimum safety distance of 50 cm can be specified. If it is determined that the distance of the vehicle to objects is below the safety distance value, a permissible path may still exist, although it will receive a lower rating than other paths for which the safety distance value is known. This reduces the probability that these paths will ultimately be selected as the movement trajectory. The advantage of this is that the determination of the movement trajectory used can be improved. A further development involves defining the vehicle's speed in the motion trajectory based on its distance to specific objects. This means that when planning the speed profile along the selected path, the speed used at each position can be provided as a function of distance. Preferably, the speed can be reduced as the distance decreases. This development offers the advantage of improved safety during motion planning. A further development stipulates that the respective state of the vehicle includes at least its orientation within its environment and a steering angle. The motion model does not generate configuration samples that are excluded by the vehicle's orientation and steering angle. However, multiple movements in forward and reverse directions, changes of direction at any point in the environment, and even steering angle changes while the vehicle is stationary can be taken into account. In particular, this prevents the generation of multiple successive configuration samples that would provide paths that are not feasible due to the vehicle's orientation and current steering angle.The configuration samples encompassing the state can build upon one another, particularly in several successive sampling steps, thereby determining which configuration samples are possible based on the preceding configuration sample. Thus, each preceding state forms the basis for a new exploration. The first state is the initial, determined state of the vehicle, and the subsequent states are determined by the configuration samples. Preferably, each configuration sample and / or state can include a possible speed of the vehicle, a distance or position relative to the preceding configuration sample, and / or a steering angle rate or steering angle change rate, which can be used to determine whether a deceleration and / or a change of direction of the vehicle's movement is necessary.For example, this can be used to determine whether steering while the vehicle is stationary is necessary for generating subsequent configuration samples by the motion model. A single-track model can preferably be used as the motion model. The single-track model provides a model of the vehicle's lateral dynamics, which can be used to describe the vehicle's movements. This further development offers the advantage of saving computing resources, which contributes to improved real-time capability. Preferably, the system checks for the possibility of passing over a low object using the height profile grid. Passing over the low object is permitted if the height and / or slope of the low object is determined to be below a specified height threshold, and if the angle of the wheel position relative to the low object is within a predefined range. This means that the movement planning can also include passing over low objects, particularly if the value of a cell in the height profile grid is below a specified height threshold and / or if at least a sequence of adjacent cells in the height profile grid are below the specified height threshold.As a further condition for crossing low objects, it can be stipulated that a wheel position is also suitable for the crossing, whereby crossings over low objects are only permitted if the wheel position is within a predefined angular range relative to the low object. In other words, the crossing can only be permitted if the wheel essentially makes direct contact with the low object. This has the advantage of allowing for further movements during motion planning. Preferably, the system provides that, for the purpose of generating a vehicle trajectory, the speed is adjusted to a predetermined value when passing over low objects. In particular, the speed can be reduced when passing over low objects. This can increase the safety of the vehicle. Furthermore, it can be stipulated that paths which do not involve crossing over low objects are given preference. In other words, the path criterion can be used to evaluate each path, with paths that involve crossing over low objects being rated lower than paths that do not. A further development proposes that areas around objects, especially tall objects, be divided into at least three distance zones: a near zone up to a first predetermined distance from the object, a middle zone extending from the first predetermined distance to a second predetermined distance from the object, where the second distance is greater than the first, and a far zone extending from the second predetermined distance to a third predetermined distance from the object, where the third distance is greater than the second. After the motion trajectory has been provided, its validity is cyclically checked, and the motion trajectory is recalculated if the environment changes.The recalculation can depend on the following criteria: If the vehicle is in the far range during the recalculation, a path and / or trajectory that runs through the far range can preferably be determined. If entering the intermediate range is unavoidable, a path and / or trajectory that leads back to the far range after entering the intermediate range can be determined. Alternatively or additionally, the speed can be reduced depending on the distance. If the vehicle is located in the central area during the recalculation, a path and / or a motion trajectory can be determined that leads away from the object and / or into the far area. If no path leading away from the object can be determined, the vehicle can preferably be stopped. If the vehicle is in close proximity during the recalculation, it is preferable to initiate a stop. In other words, an initially calculated trajectory or path may no longer be valid after a change in the environment. Therefore, a cyclical check can be performed to determine whether the planned path is still valid, and a recalculation can be carried out if it is not. If the path remains valid, it can be used. During recalculation, it is preferable to alter the existing path as little as possible and to avoid dead ends. This enhancement offers the advantage of further improving motion planning. For use cases or application situations that may arise during the procedure and are not explicitly described here, it may be provided that, according to the procedure, an error message and / or a request for user feedback is issued and / or a default setting and / or a predetermined initial state is set. Another aspect of the invention relates to a motor vehicle comprising at least one computing device, wherein the computing device is configured to perform a previously mentioned method. This offers the same advantages and possibilities for variation as the method itself. The motor vehicle according to the invention is preferably configured as a motor vehicle, in particular as a passenger car or truck, or as a passenger bus or motorcycle. The invention also includes a control device for the motor vehicle. The control device can comprise a data processing device or a processor unit configured to carry out an embodiment of the method according to the invention. For this purpose, the processor unit can comprise at least one microprocessor and / or at least one microcontroller and / or at least one FPGA (Field Programmable Gate Array) and / or at least one DSP (Digital Signal Processor). In particular, a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), or an NPU (Neural Processing Unit) can be used as the microprocessor. Furthermore, the processor unit can comprise program code configured to carry out the embodiment of the method according to the invention when executed by the processor unit. The program code can be stored in a data memory of the processor unit.The processor setup can be based on at least one circuit board and / or at least one SoC (System on Chip). As a further solution, the invention also includes a computer-readable storage medium comprising program code which, when executed by a computer or a computer network, causes it to execute an embodiment of the method according to the invention. The storage medium can be provided at least partially as a non-volatile data storage medium (e.g., as flash memory and / or as an SSD - solid state drive) and / or at least partially as a volatile data storage medium (e.g., as RAM - random access memory). The storage medium can be located within the computer or computer network. However, the storage medium can also be operated, for example, as an app store server and / or cloud server on the internet. The computer or computer network can provide a processor circuit with, for example, at least one microprocessor.The program code can be provided as binary code and / or as assembly code and / or as source code of a programming language (e.g. C) and / or as a program script (e.g. Python). The invention also includes further developments of the motor vehicle according to the invention, which have features already described in connection with the further developments of the method according to the invention. For this reason, the corresponding further developments of the motor vehicle according to the invention are not described again here. The invention also includes combinations of the features of the described embodiments. The invention therefore also includes realizations that each exhibit a combination of the features of several of the described embodiments, provided that the embodiments have not been described as mutually exclusive. The following are exemplary embodiments of the invention. Figure 1 shows a schematic representation of a motor vehicle according to an exemplary embodiment; Figure 2 shows a schematic process diagram according to an exemplary embodiment; Figure 3 shows a configuration space with configuration sample values ​​for an exemplary motion planning; Figure 4 shows a configuration space with grid cells that encompass a grid cell object spacing; Figure 5 shows a schematic representation of a vehicle passing over a low object; Figure 6 shows a representation of distance ranges around an object. The exemplary embodiments described below are preferred embodiments of the invention. In these exemplary embodiments, the described components each represent individual features of the invention, which can be considered independently of one another and each further develops the invention independently. Therefore, the disclosure is intended to include combinations of features of the embodiments other than those shown. Furthermore, the described embodiments can also be supplemented by further features of the invention already described. In the figures, identical reference symbols denote functionally equivalent elements. Figure 1 shows a schematic representation of a motor vehicle 10, which can, for example, be designed as a car. The motor vehicle 10 can be an autonomous vehicle equipped to perform motion planning to a target position, particularly for a parking maneuver, and to drive autonomously to the target position based on this motion plan. For this purpose, the motor vehicle 10 can have a sensor device 12, a computing device 14, and a control device 16. The sensor device 12 can comprise one or more sensors capable of acquiring data from the environment of the motor vehicle 10 and / or data from the motor vehicle 10 itself. For example, the sensor device 12 can comprise optical sensors, in particular camera sensors and / or laser sensors, radar sensors and / or ultrasonic sensors, with which the environment of the motor vehicle can be recorded. Furthermore, the sensor device 12 can comprise sensors capable of determining the current steering angle, speed and / or orientation of the motor vehicle 10 in its environment. The computing unit 14 and the control unit 16 can be provided as the vehicle computer of the motor vehicle 10, either as a unit in a common device or as separate devices. The computing unit 14 can be configured at least to perform motion planning to a target position in the environment of the motor vehicle 10 as determined by the sensor device 12. The motion planning can be based on a sampling-based algorithm in which a plurality of samples can be determined, each of which, as a sequence, yields a geometric path to the target position. In particular, the computing unit 14 can include a graphics processing unit 15, which enables fast parallel processing of multiple samples. The control unit 16 can be configured to control the motor vehicle 10 by means of the motion planning provided by the computer unit 14, in particular by means of a provided motion trajectory, such that the motor vehicle 10 drives autonomously to the target position. In this context, several steering maneuvers with changes of direction, including small and large steering inputs, as well as steering inputs when the motor vehicle 10 is stationary, can be taken into account by the motion trajectory. Figure 2 shows a schematic process diagram for the motion planning of an autonomous motor vehicle 10 according to an exemplary embodiment, wherein the motion planning can be carried out by means of the sensor device 12, the computing device 14 and the control device 16. In step S1, the sensor device 12 of the vehicle 10 can determine an initial state of the vehicle 10 and its environment, which includes a predetermined parking position. The environment can include static and dynamic objects in the vicinity of the vehicle 10. The initial state can provide an initial configuration or starting state of the vehicle 10 from which the motion planning is carried out. For example, the initial state can include a current speed, a current steering angle, and / or a current orientation in space or the environment. In step S2, the computing unit 14 can determine a target position of the motor vehicle, which may additionally include a target state. The target position can be the position to which the motor vehicle 10 is to travel, and the target state can include the orientation and / or speed that the motor vehicle 10 is to assume at the target position. The target position can be known, for example, due to a predetermined route, such as through GPS route planning, and / or the target position can be a parking position that the motor vehicle has determined in its surroundings. In step S3, a virtual configuration space 18 representing the environment of the motor vehicle 10 can be generated by the computing unit 14 from the environment determined by the sensor device 12. Such a configuration space 18 is shown in an exemplary and highly schematic way in Fig. 3. The computing unit 14 can then calculate configuration samples 20, 20', 20" in the configuration space 18 using a sampling-based algorithm, preferably using at least one graphics processing unit 15. These configuration samples comprise possible positions of the motor vehicle 10 with a corresponding state of the motor vehicle 10. The configuration samples 20, 20', 20" (samples) can include, as a state, at least a steering angle required for the position and an orientation of the motor vehicle at that position.In particular, the configuration samples 20, 20', 20" can be generated in several sequential sampling steps. For example, in a first sampling step, the configuration samples 20 can be determined; in a subsequent second step, the configuration samples 20', which build upon the first configuration samples 20, especially upon the state provided in the configuration samples 20, can be determined; and similarly, in a third sampling step, further configuration samples 20" can be determined. The number of sampling steps shown here is only an example, and a large number of further sampling steps can be performed up to a final sampling step in which the vehicle 10 reaches the target position. To limit the number of configuration samples in each sampling step and thus provide real-time capability for motion planning, it can further be provided that the generation of configuration samples 20, 20', 20" in configuration space 18 is carried out by a motion model, thereby excluding impermissible movements of the vehicle. In particular, the motion model can be configured so that positions and / or states of the vehicle 10 are not generated in configuration space 18 for which movement of the vehicle is excluded, while all other possible movements of the vehicle are generated, preferably also assuming a steering movement while the vehicle is stationary. A single-track model is particularly preferred as the motion model, allowing steering angles and the resulting movement and orientation of the vehicle 10 in space to be easily determined. The single-track model can preferably assume front-axle steering of the vehicle 10 and take forward and / or reverse movements into account. This allows, for example, the planning of multiple trains with changes of direction at any point. From sequences of the generated configuration sample values ​​20, 20', 20" geometric paths 22, 22' of the motor vehicle 10 can be determined, which preferably lead to the target position after the last sampling step. In step S4, the configuration space can be subdivided into several grid cells 26 in order to determine the distance to objects 24 in the vicinity for each grid cell 26 and to determine the distance of the motor vehicle 10 to the at least one object 24 at that position. Step S4 can preferably take place in parallel with step S3. This means that the grid cells 26, which encompass the distance to the respective objects 24 in the vicinity, can already be generated when creating the configuration space 18 in order to take the distance into account when generating the configuration sample values ​​(20, 20', 20"). This is illustrated, for example, in Fig. 4. Figure 4 shows the configuration space 18, which is rasterized into several grid cells 26 to provide a grid cell object distance 28 to objects 24 in the vicinity. For clarity, only the determination of the grid cell object distances 28 for one grid cell 26 is shown, although this determination can be performed for each grid cell 26. Preferably, the grid cell object distance 28 can be determined from the center of the grid cell 26 to a respective object 24, in particular to a point of the object 24 nearest the grid cell 26. Furthermore, the shape of the vehicle 10 can be modeled in the configuration space 18 as a vehicle circle model, wherein the vehicle shape in this example is modeled by means of two circles 30.Thus, for each configuration sample value 20 planned for grid cell 26, it can be checked whether the grid cell object spacing 28 provides a sufficient distance to the objects 24 by subtracting the grid cell object spacing 28 from the radius of at least one circle 30 of the vehicle circle model. If the distance calculated in this way is, for example, less than a minimum distance, a corresponding path 22 and / or a configuration sample value 20 can be excluded from motion planning. In particular, no further sample values ​​can then be generated based on this configuration sample value 20, thereby minimizing computational effort. Particularly preferably, in step S4, the objects 24 can be further classified as low or high, whereby for high objects the distance of the vehicle described above can be checked to ensure the minimum distance is maintained. For low objects 24, on the other hand, it can be checked whether a crossing is possible and / or whether a lateral distance of the wheels 32 to these low objects 24 is maintained to avoid damage to vehicle rims. In particular, the configuration space 18 can additionally provide a height profile grid, whereby height information can be provided in the respective cells of the height profile grid. Thus, preferably, a sequence of cells of the height profile grid can also provide inclination information for the low objects to check whether a crossing is possible. To verify whether a lateral distance of wheel 32 to the low object 24 is maintained, the planned wheel position at the respective location can be determined from the respective configuration scan values. From the wheel position, a lateral distance of wheel 32 to the low object 24, which could be, for example, a sidewalk, can then be determined. The corresponding configuration scan value and / or the associated path can be excluded from motion planning after the scan step if the lateral distance is below a wheel distance threshold, for example, below 10 cm. To verify whether a crossing over a low object 24 is possible, the height profile grid can be used to check whether the height and / or inclination of the low object 24 is below a height threshold and, preferably, whether a wheel position, which can be read from the configuration scan values ​​(20, 20', 20''), is within a predefined angular range relative to the low object 24. For illustration, Fig. 5 shows a schematic representation of a crossing over a low object 24. Here, a crossing of the low object 24 can be checked using the previously determined height profile grid, whereby the crossing can be permitted if the height and / or inclination of the low object 24 is below a height threshold. Furthermore, it can be checked whether the angle δ at which the at least one wheel 32 strikes the object 24 is within a predefined permissible angular range.In particular, the crossing of the low object 24 can only be permitted if the wheel 32 with a running surface hits the low object 24 essentially straight on, whereby the specified angle range may, for example, additionally allow angles of + / - 45 degrees for a crossing. In step S5, the possible paths 22, 22' and / or the newly generated configuration sample values ​​(20, 20', 20'') can be evaluated after each sample step depending on a path criterion. This path criterion can, at a minimum, check whether the calculated distance of the vehicle 10 to the respective objects in the environment meets a minimum distance requirement. The minimum distance can be defined as a minimum distance that the vehicle 10 should maintain to the respective objects, for example, 20 cm. Additionally, the path criterion can also evaluate, for example, the number of required steering maneuvers on the respective path, the number of required changes of direction, the path length, the path duration, and / or steering maneuvers while the vehicle is stationary.Preferably, only paths that maintain the minimum distance can be followed further, whereby paths and / or configuration sample values ​​below the minimum distance are excluded and cannot be followed further. Thus, the evaluation can be performed iteratively after each sampling step, and paths can be excluded if necessary, until the paths lead to the target position in a final sampling step. In step S6, at least one of the calculated paths (22, 22') can then be provided to the control unit 16 as the motion trajectory of the motor vehicle 10. Consequently, the control unit 16 can autonomously control the motor vehicle 10 to the target position using the motion trajectory. Preferably, from the evaluated paths 22, 22' to the target position, a path 22', in particular the best-evaluated path 22', can be selected for which a speed profile can be determined. Thus, a motion trajectory of the motor vehicle 10 to the target position can be generated from the geometric path. In step S7, after the motion trajectory for controlling the motor vehicle 10 has been provided, the validity of the motion trajectory can be repeatedly checked, in particular whether it remains valid after a change in the environment. If it is no longer valid, a new recalculation can preferably take place depending on the existing distance to objects 24. Figure 6 shows distance ranges A1, A2, and A3 around an object 24, which can provide criteria for recalculating the motion trajectory if a previously determined motion trajectory becomes invalid due to a change in the environment. Specifically, a near range A1, a medium range A2, and a far range A3 are shown. If the vehicle is located in the far region A3 during the recalculation, it is preferably possible to determine a new path and / or a new trajectory that is also located in the far region A3 or even further away from the object 24. However, if entering the intermediate region A2 cannot be avoided during the recalculation, the new path and / or trajectory may lead from the intermediate region A2 back into the far region A3. It is particularly preferable that the closer the vehicle 10 gets to the object, the more its speed can be reduced. If the motor vehicle is located in the central area A2 when the new calculation takes place, a new path and / or a new trajectory of movement may be provided, leading away from object 24 and preferably into the distant area A3. If no such path is available, it may preferably be provided that the speed of the motor vehicle 10 is reduced until it comes to a stop. If, during the recalculation, vehicle 10 is located in the immediate vicinity of A1, it may be stipulated that the vehicle be stopped. Preferably, a request for manual control of vehicle 10 may then be issued. Overall, the examples show how a strategy for movement planning in confined spaces can be provided.

Claims

Method for motion planning of an autonomous motor vehicle (10), wherein the method comprises the following steps: - Determining (S1) an initial state of the motor vehicle (10) and an environment of the motor vehicle (10) by a sensor device (12) of the motor vehicle (10); - Determining (S2) a target position and a target state of the motor vehicle (10) in the environment by a computing device (14) of the motor vehicle (10); - Determining (S3) configuration samples (20, 20', 20'') in a configuration space (18) by the computing device (14), wherein the configuration samples (20, 20', 20'') are generated depending on a motion model of the motor vehicle (10) and thus impermissible movements of the motor vehicle (10) are excluded, wherein the configuration space (18) represents the environment of the motor vehicle (10) and the configuration samples (20, 20',20'') possible positions of the motor vehicle (10) with a respective corresponding state of the motor vehicle (10), wherein the configuration samples (20, 20', 20'') are determined using a sampling-based algorithm in predefined sampling steps, wherein in each sampling step a plurality of configuration samples (20, 20', 20'') are determined in parallel using at least one graphics processing unit (15) of the computing device (14), wherein the configuration samples (20, 20', 20'') generated in each sampling step are based on the state and position of the configuration samples (20, 20', 20'') of the preceding sampling step, wherein respective sequences of configuration samples (20, 20', 20'') possible paths (22,22') of the motor vehicle (10) through the configuration space (18) to reach the target position and target state;- wherein the configuration space (18) is rasterized into several grid cells (26) (S4) and a grid cell object distance (28) is determined from a center of a respective grid cell (26) to one of the nearest objects (24) in the environment of the respective grid cell (26), wherein a motor vehicle shape of the motor vehicle (10) in the configuration space (18) is modeled as a motor vehicle circle model comprising a circle (30) or a composition of several circles, wherein for a respective grid cell (26) a distance of the motor vehicle (10) to the nearest object (24) in the environment is calculated by a difference of the grid cell object distance (28) and a radius of a circle (30) of the motor vehicle circle model,- evaluating (S5) the possible paths (22, 22') and / or newly generated Configuration sample values ​​(20, 20',20'') after each sampling step by the computing device (14) depending on a path criterion, wherein the path criterion checks at least the calculated distance of the motor vehicle (10) to the objects (24) in the environment for the presence of at least a minimum distance, excluding paths and / or configuration sampling values ​​for which the minimum distance is not present;- providing (S6) at least one possible path (22') as a motion trajectory of the motor vehicle to a control device (16) of the motor vehicle (10) for controlling the autonomous motor vehicle (10) to the target position in the target state;- wherein objects (24) in the environment are classified as low and high objects (24), wherein the presence of the minimum distance is checked for high objects (24), wherein for low objects (24) the configuration space (18) is subdivided into a height profile grid with slope information, wherein for configuration sampling values ​​(20, 20', 20''),which are located near a low object (24), a wheel position of the motor vehicle is determined from the state of the respective configuration scan values ​​(20, 20', 20'') in relation to the low object (24), wherein, depending on the wheel position, a lateral distance of a respective wheel (30) to the low object (24) is determined, whereby the possible path (22) and / or the configuration scan value for which the lateral distance is below a wheel distance threshold value is excluded. Method according to claim 1, wherein paths whose distance to objects is below a predetermined safety distance value are rated worse by the path criterion than paths above the safety distance value. Method according to one of the preceding claims, wherein in the motion trajectory a speed of the motor vehicle (10) is determined as a function of a distance of the motor vehicle (10) to respective objects (24). Method according to one of the preceding claims, wherein the respective state of the motor vehicle (10) comprises at least an orientation of the motor vehicle (10) in the environment and a steering angle, wherein the motion model does not generate configuration sample values ​​(20, 20', 20'') that are excluded by the orientation of the motor vehicle (10) and the steering angle. Method according to one of the preceding claims, wherein a crossing of the low object (24) is checked on the basis of the height profile grid, wherein a crossing of the low object (24) is allowed if a height and / or inclination of the low object below a height threshold value is determined from the height profile grid and if an angle (δ) of the wheel position to the low object lies within a predetermined angular range. Method according to claim 5, wherein for the creation of a motion trajectory of the motor vehicle (10) a speed is adjusted to a predetermined speed value when passing over low objects (24). Method according to one of claims 5 or 6, wherein preferred are those paths in which no crossing of low objects is provided. A method according to one of the preceding claims, wherein areas around objects (24), in particular tall objects (24), are subdivided into at least three distance ranges (A1, A2, A3), a near range (A1) up to a first predetermined distance from the object (24), a middle range (A2) extending from the first predetermined distance to a second predetermined distance from the object (24), wherein the second distance is greater than the first distance, and a far range (A3) extending from the second predetermined distance to a third predetermined distance from the object (24), wherein the third distance is greater than the second distance, wherein, after the provision of the motion trajectory, the validity of the motion trajectory is cyclically checked.and if the environment changes, the motion trajectory is recalculated using the following criteria: - if the motor vehicle (10) is in the far area (A3) during the recalculation: ◯ Determine a path and / or motion trajectory that runs in the far area (A3); ◯ if entry into the intermediate area (A2) is unavoidable, determine a path and / or motion trajectory that leads back to the far area (A3); ◯ Reduce speed depending on the distance; - if the motor vehicle (10) is in the intermediate area (A2) during the recalculation: ◯ Determine a path and / or motion trajectory that moves away from the object (24) and / or leads into the far area (A3); ◯ if no path moving away from the object is determined, stop the motor vehicle (10); - if the motor vehicle (10) is in the near area (A1) during the recalculation is located:◯ Stopping the motor vehicle (10)., Motor vehicle (10) comprising at least one computing device (14), wherein the computing device is configured to perform a method according to one of the preceding claims.

Images