JPWO2021197729A5 - - Google Patents

Description

The present invention relates to a method for planning a target trajectory according to the preamble of claim 1.

From DE 10 2015 208 790 A1 methods and systems for automatically determining trajectories for vehicles are known. The trajectory connects a starting point corresponding to the vehicle's current location to a destination point. In this method, a plurality of waypoints are determined and at least one first sub-trajectory connecting the starting point to one of the waypoints is determined. Furthermore, a plurality of second subtrajectories are determined, in each case connecting the target point to one of the waypoints. Further, the trajectory is determined by selecting one of the at least one first sub-trajectory and the second sub-trajectory, and the at least one component of the vehicle is controlled based on the determined trajectory, and the at least one component of the vehicle is controlled based on the determined trajectory, and the at least one component of the vehicle is controlled based on the determined trajectory. The subtrajectories terminate at their respective waypoints.

Furthermore, WO 2019/223909 describes a method for at least partially automated control of a motor vehicle. The method includes receiving an ambient signal representing a surrounding environment of the vehicle as sensed by an ambient sensor of the vehicle. Based on the received ambient signal, an object located in front of the vehicle with respect to the direction of travel of the vehicle is detected. Furthermore, the method provides for determining whether there is a road fork in the overtaking trajectory for overtaking an object and whether any on-coming traffic associated with the motor vehicle is blocked during overtaking. If the decision reveals that there are no road junctions within the overtaking trajectory to overtake the object and that oncoming traffic is not blocked during overtaking, the vehicle's lateral and longitudinal A control signal for at least partially automated control of the guidance is output.

The invention is based on the object of specifying an improved method for planning target trajectories to be traveled by a vehicle in an automated manner.

This object is solved according to the invention by a method with the features specified in claim 1.

Advantageous developments of the invention are the subject matter of the dependent claims.

A method for planning a target trajectory to be traveled in an automated manner by a vehicle, the method comprising: determining a plurality of distinct trajectories as candidates for the target trajectory, each of the trajectories comprising an array of multiple It is made up of 2 track sections. The method further provides that the planning is based on selecting one of the plurality of distinct trajectories as a target trajectory, and the selection is based on evaluating the trajectory according to a predefined cost function and determining the most cost-effective trajectory. Identification of highly rated trajectories. According to the invention, an array of subtrajectories each having the same position specifications and different dynamics specifications is associated with each trajectory section. In this context, position specifications are to be understood as specifications regarding the position path that the vehicle should follow when traveling along the respective track section, and dynamics specifications are specifications regarding the dynamics of the vehicle, in particular the respective track sections. It is to be understood as a specification regarding the acceleration and/or speed at which the vehicle should travel when traveling along a section. If a change is detected in the boundary conditions to be observed and/or the driving task to be performed, the sub-trajectories may be Pre- ( feedforward ) control over selection by adapting the cost function of individual subtrajectories to changed boundary conditions and/or driving tasks, such that lower cost functions are assigned to subtrajectories that are more suitable. is executed.

By applying this method, vehicle operation in an automated manner can perform different driving tasks, and the driving tasks can only be performed if safety-critical standards are not violated. can be guaranteed as much as possible.

Driving tasks include, inter alia, forming emergency lanes, reducing the driving speed of the vehicle in certain driving situations as a precaution, changing lanes for specific vehicles, e.g. police and/or emergency services, parking the vehicle on the shoulder of the road, etc. , and/or consideration of degradation of the vehicle's steering or braking system.

The method makes it possible to perform different driving tasks with target pre- control of target trajectory planning in real time. If performing a driving task would result in a violation of safety constraints, the target trajectory plan can override the specification and provide a safer target trajectory.

In one embodiment of the method, each trajectory determined as a candidate, and therefore also the target trajectory selected from the set of these candidates , is a data set in which the vehicle travels along the respective trajectory. It contains information about the positional path to be followed and further information about the dynamics , in particular about the acceleration and/or driving speed at which the vehicle should travel when traveling along the respective trajectory. The target trajectory selected from multiple trajectories not only determines along which position coordinates the vehicle should travel in autonomous driving mode, but also determines how the vehicle should move dynamically, i.e. , it is specified at what point in time the vehicle should be at each location coordinate. The method thus makes it possible to find the optimal position path for automatic vehicle guidance and at the same time to find the optimal vehicle dynamics.

Furthermore, in another embodiment, the plurality of distinct trajectories from which the target trajectory is selected is determined by determining predefined trajectory support points as possible vehicle positions in the forward region. by selecting from a set of trajectory support points a plurality of array points traveling in the direction of travel and determining sub-trajectories each passing through one of the selected array points; Discretized. In other words, the track support points represent positions in the forward area, in each case guided by one or more of the tracks. Each trajectory is thus guided through a predefined trajectory support point, the sections between two trajectory support points forming the above-mentioned trajectory sections, and each of these sections having the above-mentioned arrangement of sub-trajectories. Associated. The individual tracks are therefore made up of sub-tracks, each of which is connected to one another at a track support point. Since the number of sub-orbitals is limited, the number of orbits composed of them is also limited. Hereinafter, this plurality of orbits will be referred to as an orbit array. Since the planning of the target trajectory is based on the selection of trajectories from the trajectory array, the target trajectory can be planned with little computational cost.

In one possible development, the cost is determined for each trajectory of the trajectory array using a predefined cost function, and the cost functions for the individual trajectory sections and the subtrajectories associated with each of them are adhered to. defined as a function of the boundary conditions to be performed or the driving task to be performed. The cost function therefore determines, for example, that the target trajectory to be selected must not depart from the lane of vehicle operation in an automated manner, that the target trajectory can be physically realized for vehicle operation in an automated manner, etc. Take into account boundary conditions.

In one possible development of the method, different cost functions defined for different boundary conditions or driving tasks are respectively predefined for different sub-trajectories of the individual track sections. These cost functions indicate how well each boundary condition or driving task can be met with respect to the target trajectory. Relatively good performance is rewarded with low costs , and relatively poor performance is punished with high costs .

Advantageously, the total cost is determined for each of the track sections by summing in a weighted manner the costs associated with the subtrajectories of the respective track section.

Advantageously, the cost of a track is determined by summing the total costs of its track sections.

In another possible embodiment, in order to determine the relatively best target trajectory, the total cost of the trajectory section is determined by a weighted sum of the costs determined for the different boundary conditions of the trajectory section. .

Subsequently, in one possible embodiment, the cost of a trajectory is determined by summing the total cost of its trajectory sections, and from the trajectory arrangement with the lowest cost, the trajectory is determined taking into account boundary conditions and/or driving tasks. selected as the target trajectory for the vehicle to drive in an automated manner.

Furthermore, in the method, the cost function is modified by a pre -control, which adapts the trajectory plan to the current driving task, and in the case of multiple driving tasks, a prioritization is performed . The purpose of pre- control is to adapt the planning of trajectories, in particular the selection of target trajectories , to the current driving task, including the required boundary conditions and , in the case of multiple driving tasks, to perform prioritization. It is to be. The target trajectory is therefore selected by considering the current driving task or, if appropriate, a plurality of compatible driving tasks to be considered.

Exemplary embodiments of the invention are explained in more detail below with the aid of the drawings.

FIG. 1 schematically shows the first step for determining the trajectory alignment. FIG. 2 schematically shows the second step for determining the trajectory alignment. FIG. 3 schematically shows the third step for determining the trajectory alignment. FIG. 4 schematically shows the cost function. FIG. 5 schematically shows the modification of the cost function by pre- control. FIG. 6 schematically shows further modification of the cost function by pre- control.

Mutually corresponding parts are provided with the same reference symbols in all figures.

FIG. 1 shows a first step for determining a trajectory arrangement in which the trajectory T shown in particular in FIG. 2 is selected as the target trajectory T _Soll , which is shown in more detail in FIG.

The vehicle 1 has an assist system for the automatic driving mode, and the corresponding sensor system continuously detects signals during the automatic driving mode.

During the automatic driving mode of the vehicle 1, the vehicle 1 is required to behave appropriately in a wide variety of driving situations and perform different driving tasks. In normal autonomous driving mode, these driving tasks include, for example, keeping distance in the center of the lane, maintaining a set speed, etc., and specialized driving tasks are, for example, as shown in Figure 5. This is understood to mean forming an emergency lane R in which the vehicle 1 is located, or avoiding a collision, for example because an obstacle is suddenly detected in the lane F of the vehicle 1.

In order to detect a wide variety of driving situations, the sensor system comprises a number of sensors arranged in and/or on the vehicle 1, which can e.g. and/or optionally integrated to extend or optimize detection range.

In order to be able to control a large number of possible driving situations during the autonomous driving mode of the vehicle 1, a trajectory planning model is usually used, which selects the best trajectory T from a large number of possible trajectories T, or Choose either to calculate the optimal trajectory T using the structuring procedure.

These two methods are based on the evaluation of the trajectory T by a cost function K, shown in particular in FIGS. 4 to 6, where the costs are composed of different partial costs with different weightings.

An example of an individual cost function K is as follows.
- deviation from the desired path;
- high level dynamics of the vehicle in longitudinal and lateral directions of vehicle 1;
-below a safe distance;
- a collision with obstacle 2, which is shown in more detail in Figure 3;
- Failure to comply with the set speed.

To ensure a relatively safe autonomous driving mode of the vehicle 1, safety-critical costs are given higher weight than costs resulting from an uncomfortable ride.

When choosing a trajectory T, it is necessary to additionally observe the so-called strict boundary conditions, in which the trajectory T is not allowed to leave the lane and it must be possible to physically realize the trajectory T. .

In order to control a plurality of different specialized driving tasks, the method described below is provided, in which trajectory planning is controlled by changing the target state of the vehicle 1 and the operated variable range.

Trajectory planning is then carried out in the absence of a higher fundamental objective in the process, for example following a safe distance before a collision occurs, an unintentional departure from lane F, excessive vehicle reaction, or inability to drive. Plan pre- control specifications .

If, for example, a collision between the vehicle 1 and the obstacle 2 is about to occur due to the sudden appearance of the obstacle 2 in front of the vehicle 1, avoiding the collision takes priority over the driving task. If such a critical situation no longer exists, the desired driving task is prioritized again .

The method determines the desired deviation of the vehicle 1 relative to the center of the lane F by the maximum permissible driving speed v _EGO and also in relation to a predefined distance and/or time. Provide continuous specifications for trajectory planning in relation to time, adjustable deceleration, permissible acceleration and adjustable steering dynamics.

Additionally, the method includes prioritizing potentially conflicting driving tasks. For example, the system for autonomous driving mode of vehicle 1 can request safe parking of vehicle 1, while at the same time so-called Move-Over-Law situations requiring different decelerations. Situation) exists .

Behavior in move-over-low situations, i.e. when evasive action rules are applied, e.g. when an emergency vehicle, e.g. a police car, is approaching, different from that required to form emergency lane R It is also possible to request an offset into lane F for vehicle 1.

In any case, with respect to the positioning of the vehicle 1 in the lane F, the driving task is prioritized by the selection of the driving speed and/or offset specification.

Furthermore, the method uses specific driving tasks required to change the specification of the cost function K in trajectory planning, such as:
- Safely parking the vehicle 1 on the road shoulder S illustrated in Figure 4;
- lane change,
- Preventive reduction of driving speed v _EGO in uncertain driving situations, e.g. wrong-way drivers in adjacent lanes or pedestrians on the road;
- Simultaneously with the reduction of the current driving speed v _EGO , providing for driving to the edge of lane F if a move-over-low situation exists.

More complex driving tasks, such as safely parking the vehicle 1 on the shoulder S or changing multiple lanes, are sent to the trajectory planning system by a time sequence of driving speed and offset specifications, so-called lane offset specifications.

If a degraded state of the brake or steering system is reported, adjustable brake or steering dynamics are adapted to the trajectory planning.

Furthermore, the adjustable deceleration of the vehicle 1, ie the reduction of the current driving speed v _EGO , is adapted to the current weather conditions.

In particular, the method provides that the target trajectory T _Soll illustrated in FIG. 3 is to be traveled by the vehicle 1, in particular in automatic driving mode, in particular without a driver.

Such a target trajectory T _Soll includes information regarding the position route that the vehicle 1 should follow when traveling along the target trajectory T _Soll , that is, information regarding the position coordinates, and includes information regarding the position coordinates of the vehicle 1 when traveling along the target trajectory T _Soll . is understood to mean a data set containing information about the acceleration or driving speed v _EGO to be moved. The target trajectory T _Soll therefore not only specifies in which position coordinates the vehicle 1 should move, but also specifies the time during which the vehicle 1 is in each position coordinate.

The planning is in this case based on the determination of several distinct candidates for the target trajectory T _Soll , the selection being based on the cost function K known from the prior art, as described above.

Figure 1 shows the coordinate system in detail, where the x-coordinates x _i to x ₄ are plotted on the x-axis in the longitudinal direction of the vehicle, i.e. in the direction of travel of the autonomous vehicle 1, and the y-coordinates y-1 to y1 are plotted on the y It is plotted on the axis and shows the lateral direction of the vehicle. Δx, ie the distance between two x-coordinates, represents a function of the current driving speed v _EGO of the vehicle 1.

Additionally, a plurality of track support points P _0,0 to P _4,2 are shown, track support point P _0,0 representing the departure point of vehicle 1, track support points P _4,0 to P _4,2 representing the vehicle Represents the destination coordinates of 1. In particular, the x coordinates x ₀ to x ₄ and the y coordinates y-1 to y1 are the xy coordinates of the orbital support point P = (x _i ,y _j ).

The track support points P _0,0 to P _4,2 are distributed in the forward region in the x-axis direction, that is, in the direction of travel of the vehicle 1. This forward area V defines the route that the vehicle 1 passes, ie travels, at the current driving speed v _EGO within a predefined time interval, for example 30 seconds. Within the forward region V, there are theoretically an infinite number of trajectories from which the target trajectory T _Soll can be selected. In order to minimize the computational power required when selecting the most suitable target trajectory _T for the vehicle 1 and the driving situation, the trajectories from which the target trajectory _T is to be selected are kept separate . For this, a set of predetermined trajectory support points P _0,0 to P _4,2 is determined in the forward region V, a plurality of trajectories T are determined, and a set of trajectory T Each is in the direction of travel through a series of trajectory support points P _0,0 to P _4,2 , as shown in the second step of Fig. 2, in order to determine the trajectory alignment in a further coordinate system. .

This multiple trajectories T form a trajectory array, and the trajectories T form the coordinates for the selection of the target trajectory T _Soll , i.e. only this limited number of trajectories T for the selection of the target trajectory T _Soll be considered.

In particular, as shown in Fig. 2, the individual track support points P _0,0 to P _4,2 are connected to each other in pairs in the x-direction, i.e. in the direction of the longitudinal axis of the vehicle, according to the order of the direction of travel. It is connected. Thus, individual trajectory sections TR, TR _(0,0)(1,1) of the trajectory T are formed, which are mapped to the selected set of target trajectories T _Soll shown in Fig. 3. .

Each trajectory section TR, TR _(0,0)(1,1) itself comprises an array of subtrajectories, not shown in more detail, each having the same xy path but different accelerations and/or velocities. Costs are associated with subtrajectories, and costs are associated using predefined cost functions K for different boundary conditions.

If a change in the boundary conditions to be observed or the driving task to be performed is detected, the cost function K for each subtrajectory of the individual trajectory section TR, TR _(0,0)(1,1) Adaptation performs pre- control of selection. This adaptation is more suitable than others for satisfying modified boundary conditions and/or performing modified driving tasks, TR _(0,0)(1,1) This is done to allocate lower costs.

For each trajectory T from the trajectory array, the cost is determined by a predetermined cost function K, which is defined _as How well each boundary condition is satisfied on each trajectory interval TR _(0,0)(1,1) along with each subtrajectory defined for and for predefined boundary conditions. Indicates whether there is a

Relatively good performance is rewarded with low costs, while relatively poor performance is punished with high costs. The sum of the costs, especially the weights, of the trajectory section _{TR (0,0)(1,1)} , determined for the various boundary conditions of the subtrajectories of the trajectory section TR _(0,0)(1,1). Based on the sum of the assigned costs, the total cost of the track section TR _(0,0)(1,1) is determined.

The cost of a trajectory T in the trajectory array is formed by the sum of the total costs of the trajectory sections TR _(0,0)(1,1) of each trajectory T. The lowest cost trajectory T is then selected from the trajectory array as the target trajectory T _Soll , as shown in FIG.

The target trajectory T _Soll selected from the trajectory array indicates the travel path of the vehicle 1 due to an obstacle 2 located in the lane F, which obstacle is detected in the forward area V.

Track sections TR that result in collisions between vehicle 1 and obstacle 2 are sanctioned by increasing their costs . This gives these trajectory sections TR a lower priority for the selection of the target trajectory T _Soll than other trajectory sections TR.

4 to 6 each show an example of a cost function.

Figure 4 shows the cost function K(y) for the lateral position of vehicle 1, including vehicle 1's lane F, left lane F1, right lane F2, lane marking M, each shoulder S or roadside strip, and the track. Support points P _i,j = (x _i ,y _j ) are shown.

Vehicle 1 driving in the middle of lane F is associated with a lower cost than if vehicle 1 were not driving in the middle. In other words, driving in the center of lane F is rewarded with a lower cost.

Vehicle 1 driving on lane marking M is sanctioned with high costs , driving in the middle of left lane F1 or right lane F2 is punished more severely than driving in the middle of lane F , and driving on lane marking M It's less punishable than driving over. Driving on the shoulder S or roadside strip is relatively severely and heavily penalized due to the corresponding high costs.

FIG. 5 first shows the progression of the cost function K(y) shown in FIG. 4 and the cost function K1(y) modified by the pre- control and its progression.

The purpose of pre- control is to adapt the trajectory plan for the selection of the target trajectory T _Soll to the required driving tasks and the necessary boundary conditions, and to prioritize if there are several driving tasks to be performed. It is for the purpose of execution. The target trajectory T _Soll is therefore selected by taking into account the current driving task or, if appropriate, a number of current driving tasks to be taken into account.

According to the exemplary embodiment illustrated in FIG. is rewarded by significantly lower costs than driving in the center of lanes F, F1, and F2.

Driving in the right lane F2 is rewarded more with a lower cost, and driving in the emergency lane R is punished with a higher cost.

A further exemplary embodiment of pre- control is shown in FIG. 6, where a cost function K(y) and a modified further cost function K2(y) are shown.

Driving in the right lane F2 is relatively highly penalized, for example because there is an accident in the right lane F2, and driving in lane F with vehicle 1 puts rescue workers on the scene of an accident at risk. They will be similarly punished for not having one.

In an analog manner, it is also possible to modify the predefined cost function K for other driving tasks and/or for other boundary conditions to be observed. The modification achieves pre- control for trajectory selection.

Driving tasks for vehicle 1 include e.g. an additional offset to the center of the corresponding lanes F, F1, F2, e.g. to form an emergency lane R, or a lorry, tunnel wall, pier, guide Increased lateral distances from objects of a particular class, such as walls, may be required to be observed.

Furthermore, the driving task, as mentioned above, requires that certain maximum permissible speeds be observed, longitudinal dynamics, in particular adjustable decelerations or permissible accelerations, or lateral dynamics, in particular yaw rate, steering angular velocity, and/or the steering dynamics in the form of lateral acceleration can be predefined as a function of the circumstances, e.g. weather conditions, driving speed v _EGO , road curvature and/or deterioration of the steering or braking system of the vehicle 1. Limited to certain values that can be used.

Furthermore, as a driving task to park the vehicle 1 on the shoulder S, for example, if the steering or braking system deteriorates, e.g. in an accident situation, the police, emergency services, pedestrians on the road, the corresponding adjacent lane F , F1 A wrong-way vehicle in F2 may require a preventive reduction in driving speed.

Furthermore, the driving task is to select a specific lane F, F1, F2, for example, the lane F, F1, F2 next to a parked police vehicle or emergency vehicle at the crime scene, i.e. in the case of a so-called move-over-low situation. , pedestrians, or vehicles driving the wrong way can be specified to be avoided.

Furthermore, the driving tasks considered for vehicle 1 are vehicle 1 performing lane changes, e.g. avoiding lanes F, F1, F2, avoiding obstacles 2, overtaking relatively slow road users. It may be to control the vehicle 1 to turn into or exit a lane .

Claims (9)

A method for planning a target trajectory (T _Soll ) to be traveled in an automated manner by a vehicle (1) having an assist system for an automatic driving mode, the method comprising:
- a plurality of distinct trajectories (T ) are determined as candidates for the target trajectory (T _Soll );
- each of said trajectories (T) is composed of a plurality of arranged trajectories (TR, TR _(0,0)(1,1) );
- said planning is based on the selection of one of said plurality of distinct trajectories (T) as a target trajectory (T _Soll );
- the selection is based on an evaluation of the trajectories (T) according to a predefined cost function (K) and on the identification of the trajectory (T) that has been evaluated as the most cost-effective;
- Each trajectory section (TR, TR _(0,0)(1,1) ) is associated with multiple subtrajectories, each with the same position specifications and different dynamics specifications, and each subtrajectory has a A cost according to a defined cost function (K) is associated,
- If a change is detected in the boundary conditions to be observed or in the driving task to be performed, the sub-trajectories may be more suitable than other sub-trajectories to comply with the changed boundary conditions or to perform the changed driving task. Pre- control over said selection is performed by adapting the cost function (K) of the individual sub-trajectories to the changed boundary conditions or driving tasks, such that the sub -trajectories that have been selected are assigned lower cost functions ;
A target trajectory planning method, characterized in that the process is executed by the assist system . The method according to claim 1,
Each trajectory (T) contains information about the positional route that the vehicle (1) should follow when traveling along the respective trajectory (T) as a data set, and each trajectory ( _T Target trajectory planning method, characterized in that it also includes further information about the dynamics that the vehicle (1) should follow when The method according to claim 1 or 2,
The plurality of distinct trajectories (T ) from which the target trajectories (T _Soll ) are selected include predefined trajectory support points (Pi,j) of the vehicle (1) in a possible forward region (V). by selecting a plurality of array points traveling in the direction of travel from a set of trajectory support points (Pi ,j ), and by determining one of said selected array points as the position of A target trajectory planning method, characterized in that discretization is performed by determining the sub-trajectories (T) such that each sub-trajectory (T) passes through the target trajectory. The method according to any one of claims 1 to 3,
A target trajectory planning method, characterized in that a cost is determined for each trajectory (T) using a predefined cost function (K). The method according to any one of claims 1 to 4,
Associated with each individual track section (TR, TR _(0,0)(1,1) ) of the plurality of track sections (TR, TR _(0,0)(1,1) ) and the corresponding track section A method for planning a target trajectory , characterized in that a cost function (K) for the sub-trajectories determined is defined in advance as a function of boundary conditions to be observed or driving tasks to be executed. The method according to any one of claims 1 to 5,
The costs associated with the subtrajectories of a track section (TR, TR _(0,0)(1,1) ) determine the total cost of each track section (TR, TR _{(0,0)(1,1) )} A target trajectory planning method, characterized in that weighted summation is performed. 7. The method according to claim 6 ,
A target trajectory planning method, characterized in that the cost of the trajectory (T) is determined by summing the total costs of the trajectory sections (TR, TR _(0,0)(1,1) ). The method according to any one of claims 4 to 7,
A target trajectory planning method, characterized in that the trajectory (T) with the lowest cost among the plurality of distinct trajectories ( T) is selected as the target trajectory (T _Soll ). The method according to any one of claims 1 to 8,
The cost function (K) is modified by the pre- control, and the pre -control adapts the plan to the current driving task and prioritizes the driving tasks to be performed if there are multiple driving tasks. A target trajectory planning method characterized by: