DE102018217775A1 - A method for preparing and performing a maneuver planning of at least one vehicle - Google Patents

Abstract

Method for preparing a maneuver planning of at least one vehicle (10, 20), comprising the steps:
- Determining trajectory data of a first vehicle (10) in the form of a defined number of spatiotemporal parameters of the first vehicle (10), each with associated probability distribution, wherein for defined times each areas with associated probability (p ₁₀ ) are determined; and
- Transferring the trajectory data by means of vehicle communication to at least one other vehicle (20).

Description

The invention relates to a method for preparing a maneuver planning of at least one vehicle. The invention relates to a method for carrying out a maneuver planning of at least one vehicle. The invention further relates to a device for preparing a maneuver planning of at least one vehicle. The invention further relates to an apparatus for performing a maneuver planning of at least one vehicle. The invention further relates to a computer program product.

State of the art

While classic systems that are to enable automated procedures must plan vehicle trajectories only for the ego vehicle, taking into account behavior of other road users, there is an effort to use vehicle to X (V2X) communication to realize cooperative maneuver planning and implementation. V2X communication includes both vehicle-to-vehicle (V2V) communication and vehicle-to-infrastructure (V21) communication between a vehicle and an infrastructure unit. The goal is to increase safety and comfort and reduce fuel consumption. An initial implementation of this vision is to be investigated and implemented, for example, in the context of the publicly funded project IMAGinE ("Intelligent maneuver automation cooperative risk prevention in real time").

Known in the prior art are cooperative maneuvering, planning and implementation. For example, in WO 17029096 A1 A method of exchanging trajectories for the next 0 to 10 seconds between vehicles is described. This exchange should make it possible to draw conclusions about the driving intention and, if necessary, to adapt the driver's own maneuver.

In DE 10 2012 011 994 A1 It describes how an exchange of the current vehicle trajectory (so-called plant trajectory), as well as the intended trajectory (so-called desired trajectory), can be used via V2X to check whether it is possible to traverse the desired trajectory.

This idea will be in WO 2017/076593 A1 extended by a procedure for the decentralized coordination of driving maneuvers between vehicles. If a collision arises between the desired trajectory of one vehicle and the plant trajectory of the other vehicle, a cooperation can be initiated in which the vehicle whose plant trajectory collides with the desired trajectory of another vehicle adapts its own plant trajectory, so that the desired trajectory of the other vehicle can become his plant trajectory.

A disadvantage of in WO 2017/076593 A1 described method is that an efficient implementation provides an exchange of messages between vehicles only in a relative distance of about 50 m, which is why a cooperative maneuvering is limited to a few limited scenarios. In addition, longer prediction times for trajectories greater than about 2 seconds to about 5 seconds in mixed traffic with non-automated vehicles are currently not reliably realizable.

In addition, approaches are known in which for the purpose of a maneuver planning a distribution of possible trajectories of other vehicles are taken into account in their own maneuver planning, such as in DE 10 2016 113 903 A1 disclosed.

Disclosure of the invention

It is an object of the present invention to provide an improved method of maneuver planning for vehicles.

• - Ermitteln von Trajektoriendaten eines ersten Fahrzeugs in Form einer definierten Anzahl von raum-zeitlichen Parametern des ersten Fahrzeugs mit jeweils zugeordneter Wahrscheinlichkeitsverteilung, wobei für definierte Zeitpunkte jeweils Bereiche mit zugeordneter Wahrscheinlichkeit ermittelt werden; und
• - Übertragen der Trajektoriendaten mittels Fahrzeug-Kommunikation an wenigstens ein weiteres Fahrzeug.

The object is achieved according to a first aspect with a method for preparing a maneuver planning of at least one vehicle, comprising the steps:

• - Determining trajectory data of a first vehicle in the form of a defined number of space-time parameters of the first vehicle, each with associated probability distribution, wherein for defined times each areas are determined with an associated probability; and
• - Transferring the trajectory data by means of vehicle communication to at least one other vehicle.

• - Empfangen von Trajektoriendaten eines ersten Fahrzeugs in Form einer definierten Anzahl von raum-zeitlichen Parametern des ersten Fahrzeugs mit jeweils zugeordneter Wahrscheinlichkeitsverteilung, wobei für definierte Zeitpunkte jeweils Bereiche mit zugeordneter Wahrscheinlichkeit ermittelt werden, durch ein zweites Fahrzeug; und
• - Verwenden der Trajektoriendaten des ersten Fahrzeugs zum Durchführen einer Manöverplanung des zweiten Fahrzeugs.

The object is achieved according to a second aspect with a method for carrying out a maneuver planning of at least one vehicle, comprising the steps:

• - Receiving trajectory data of a first vehicle in the form of a defined number of spatiotemporal parameters of the first vehicle, each with associated probability distribution, wherein for defined times each areas are determined with an associated probability, by a second vehicle; and
• Use the trajectory data of the first vehicle to perform a maneuver planning of the second vehicle.

As a result, a maneuver planning of a vehicle can be improved. This is accomplished by performing a trajectory determination in a non-deterministic fashion by not defining the trajectory as a list of states, eg, positions or poses (position plus orientation), but also velocity, acceleration, etc., but is determined as a list of probability distributions. By transmitting the trajectory data to at least one adjacent vehicle, improved maneuver planning is enabled for the adjacent vehicle.

• - eine Ermittlungseinrichtung zum Ermitteln von Trajektoriendaten eines ersten Fahrzeugs, wobei die Trajektoriendaten als eine definierte Anzahl von raum-zeitlichen Aufenthaltsbereichen des ersten Fahrzeugs mit einer zugeordneten Wahrscheinlichkeitsverteilung erstellt werden, wobei für definierte Zeitpunkte jeweils Bereiche pro zugeordneter Wahrscheinlichkeit ermittelt werden; und
• - eine Übertragungseinrichtung zum Übertragen der Trajektoriendaten per Fahrzeug-Kommunikation an wenigstens ein weiteres Fahrzeug.

According to a third aspect, the object is achieved with a device for preparing a maneuver planning of at least one vehicle, comprising:

• a determination device for determining trajectory data of a first vehicle, wherein the trajectory data are generated as a defined number of space-time location areas of the first vehicle with an associated probability distribution, wherein for defined points in time areas are determined per assigned probability; and
• a transmission device for transmitting the trajectory data by vehicle communication to at least one further vehicle.

• - eine Empfangseinrichtung zum Empfangen von Trajektoriendaten eines ersten Fahrzeugs, wobei die Trajektorie als eine definierte Anzahl von raum-zeitlichen Aufenthaltsbereichen des ersten Fahrzeugs mit einer zugeordneten Wahrscheinlichkeitsverteilung erstellt werden, wobei für definierte Zeitpunkte jeweils Bereiche pro zugeordneter Wahrscheinlichkeit ermittelt werden; und
• - eine Planungseinrichtung zum Planen der Manöverplanung des Fahrzeugs mittels der empfangenen Trajektoriendaten des ersten Fahrzeugs.

According to a fourth aspect, the object is achieved with a device for carrying out a maneuver planning of at least one vehicle, comprising:

• a receiving device for receiving trajectory data of a first vehicle, wherein the trajectory are created as a defined number of space-time location areas of the first vehicle with an associated probability distribution, wherein for defined points in time areas are determined per assigned probability; and
• a planning device for planning the maneuver planning of the vehicle by means of the received trajectory data of the first vehicle.

Advantageous developments of the method are the subject of dependent claims.

An advantageous development of the method provides that the trajectory data comprise at least one of the following state profiles of the first vehicle over time: position, orientation, speed, acceleration. The trajectories actually represent state curves over time. It is intended to use a distribution of the state in the state space instead of a concrete, deterministic state. As a result, the velocity vector is not a vector per se, but a distribution over the vectors, which advantageously supports that intermediate positions of the vehicles can be calculated even better at defined times.

A further advantageous development of the method provides that the trajectory data are displayed according to a Gaussian mixing distribution in the longitudinal and transverse directions. As a result, in particular a representation of the vehicles in transverse positions or in curve areas can be better realized.

A further advantageous development of the method provides that the trajectory data comprise a matrix of occupancy probabilities of cells into which the environment is divided. The cells are not necessarily uniform. In this way, it is advantageously supported that collision probabilities can be calculated more efficiently, because algorithms known per se can be used for this purpose.

A further advantageous development of the method is characterized in that a defined collision probability of the first vehicle with the at least one further vehicle is taken into account for determining the trajectory data. As a result, the trajectories of the vehicles can be determined taking into account a defined collision probability.

A further advantageous development of the method provides that a mutual replacement of the trajectory data between the first vehicle and the at least one further vehicle is performed. In this way, a maneuver planning for the two vehicles can be performed even better.

As a result, a cooperative maneuver planning between the first vehicle and the at least one further vehicle can be carried out in a simple manner. By means of the cooperative maneuver planning, trajectory data is exchanged between vehicles, whereby a traffic scenario with several vehicles is advantageously optimized.

The invention will be described in detail below with further features and advantages with reference to several figures. All described or illustrated features, alone or in any combination form the subject matter of the invention, regardless of their summary in the claims or their Relation, as well as regardless of their formulation or representation in the description or in the figures.

Disclosed features and advantages of the device result in an analogous manner from apparent features and advantages of the method and vice versa.

• 1 eine prinzipielle Darstellung einer ersten Variante zur Ermittlung von Trajektorien eines Fahrzeugs;
• 2 eine prinzipielle Darstellung einer zweiten Variante zur Ermittlung von Trajektorien eines Fahrzeugs;
• 3 eine prinzipielle Darstellung einer dritten Variante zur Ermittlung von Trajektorien eines Fahrzeugs;
• 4 eine prinzipielle Darstellung einer kooperativen Manöverplanung zweier Fahrzeuge;
• 5 ein prinzipielles Blockschaltbild einer Vorrichtung zum Vorbereiten einer Manöverplanung wenigstens eines Fahrzeugs;
• 6 ein prinzipielles Blockschaltbild einer Vorrichtung zum Durchführen einer Manöverplanung wenigstens eines Fahrzeugs; und
• 7 einen prinzipiellen Ablauf einer Ausführungsform des vorgeschlagenen Verfahrens zum Vorbereiten einer Manöverplanung wenigstens eines Fahrzeugs.

In the figures shows:

• 1 a schematic representation of a first variant for determining trajectories of a vehicle;
• 2 a schematic representation of a second variant for determining trajectories of a vehicle;
• 3 a schematic representation of a third variant for determining trajectories of a vehicle;
• 4 a schematic representation of a cooperative maneuver planning of two vehicles;
• 5 a schematic block diagram of an apparatus for preparing a maneuver planning of at least one vehicle;
• 6 a schematic block diagram of an apparatus for performing a maneuver planning of at least one vehicle; and
• 7 a basic sequence of an embodiment of the proposed method for preparing a maneuver planning of at least one vehicle.

Description of embodiments

In particular, it is an object of the invention to provide an improved method and apparatus for performing maneuver planning of at least one vehicle.

A method is proposed which represents vehicle trajectories as a list of space-time areas (position and orientation, and optionally speeds, over time) with probability distributions and data relating to these vehicle trajectories via a wireless communication (eg V2X) to at least one other vehicle in the vicinity of Ego vehicle transmitted. This allows the replacement of arbitrarily long trajectories between the vehicles to tune decentralized cooperative driving maneuvers within a group of vehicles.

The proposed method includes cooperative maneuvering within a group of at least two or more vehicles using V2V or V2X communication, respectively. The maneuvering takes place via the exchange of trajectories between adjacent vehicles. The vehicles use the data obtained from the trajectories to plan a trajectory for themselves ("ego trajectory") that does not collide with the trajectories of the other vehicles. An important task in maneuver planning is to calculate collision-free trajectory pairs (consisting of ego trajectory and trajectories of other vehicles).

It is proposed to display the trajectories as a time-ordered list of spatio-temporal areas (position and orientation, and optionally speeds, over time) with their respective probability distributions and to send by wireless communication instead of a list of time-position points, as in known standard approaches according to WO 2017/029096 A1 and WO 2017/076593 A1 ,

The vehicle then calculates a collision probability as a function of the probability distributions of all received trajectories, wherein the collision probability of the ego trajectory with the others is calculated in the planning. This allows maneuver planning to be formulated as an optimization problem in which the collision probability is minimized, for example. As a rule, a cost function is minimized / reward function maximized and compliance with a limit of collision probability (eg collision probability <0.001) is required.

This approach offers several advantages over the standard approaches of the prior art.

First, a collision probability threshold may be used to detect when the trajectory calculation was unsuccessful, so that an emergency mode for the vehicle may be initiated. With this approach, a valid ego trajectory can be found (for example, with a very low collision probability) that would be discarded with a standard approach because it is not completely collision free.

With uncertainties in the state estimation (localization and other quantities), there are actually no deterministic trajectories. Neglecting the uncertainties in planning and planning deterministically, it can happen that a collision is produced because the planned deterministic trajectory is too optimistic and exactly complies with the safety margin, whereas a collision probability planned trajectory inflates it. Furthermore, with the proposed approach, an insecure localization of vehicles can be uniformly mapped.

Finally, the probability-based calculation of the ego trajectory leads to a stable result of the maneuvering adjustment and avoids "swinging" between different calculated trajectories.

An implementation of the maneuvering can, for example, by means of in WO 2017/076593 A1 be made described method. For this reason, the proposed method can be regarded as an improvement of this method described therein.

The proposed method will be described below with reference to a ramp-up scenario on a highway or expressway. It is considered a scenario in which a second vehicle 20 , the so-called "vehicle on the main carriageway" on the rightful lane of a highway or a highway drives on which a first vehicle 10 ("Threader") would like to ascend. Proposed is the cooperative maneuvering between the vehicles 10 . 20 from the perspective of the first vehicle 10 to implement as follows:

The first vehicle 10 calculates his ego trajectory and sends it via V2X communication to the second vehicle 20 , A pure open-loop prediction of the state along a reference trajectory would make the probability density representing the state increasingly broader. The first vehicle 10 so it would be increasingly uncertain about its position, speed, etc. In this case, the information about these quantities would be increasingly imprecise towards the end of the trajectory and would be irrelevant. In real systems, the first vehicle can 10 localize at later times and therefore specify its state over a closed-loop prediction again and again, such as from A. Bry and N. Roy, "Rapidly Exploring Random Trees for Motion Planning and Uncertainty", IEEE International Conference on Robotics and Automation (ICRA), 2011 known.

This can be the first vehicle 10 You may want to use a location map that tells you how to locate and where. If one proceeds in this way, the probability density, which represents the state, remains realistic in the course of the entire trajectory, so that the evaluation of the probability of collision also yields reasonable values. For the representation of the trajectory of the first vehicle 10 (Ego trajectory) are several variants conceivable, which are explained below.

1 shows a first variant of a representation of the Ego trajectory by the first vehicle 10 , where the first vehicle 10 For example, for two different times t = 2s and t = 5s local areas or positions determined by the first vehicle 10 with a defined probability p ₁₀ be occupied. As a result, a matrix of spatial regions T (t, t, p ₁₀ ) for several times t and several probabilities p ₁₀ created.

Recognizable are two dashed and dotted areas for the times t = 2 s and t = 5s. The borders represent positions of the first vehicle 10 which are taken with defined probability values. Recognizable are positions of the first vehicle 10 , which at the two times t = 2s and t = 5s with a probability p ₁₀ 90% (dotted line) or 99% (dotted line). The illustrated areas correspond to physical dimensions of the entire first vehicle 10 , In this way, collision checks for collisions of the first vehicle 10 with the second vehicle 20 relieved, since the respective vehicle dimensions do not need to be considered additionally.

In a second variant, the in 2 is shown, becomes a list of positions of the first vehicle 10 determined with a multivariant Gaussian distribution (in longitudinal and transverse direction) per time step. Recognizable are the two multivariable Gaussian distributions p ₁₀ at the two times t = 2s and t = 5s. In this way, in particular cases can be mapped, in which the first vehicle 10 located in a curve and can locate better transverse than longitudinal.

A third variant of the creation of the trajectory of the first vehicle 10 is in 3 in this case, per time step, a list of occupancy probabilities p ₁₀ is created in a grid in Euclidean space. This has the advantage of having simple scheduling algorithms such as A * (for calculating a shortest path between two nodes in a graph) for the maneuver planning of the two vehicles 10 . 20 can be used. Similar to the first variant of 1 correspond to the areas shown p ₁₀ Extensions of the entire two vehicles 10 . 20 ,

Optionally, a distribution over the velocity vector per time point v (t) in the trajectory can be contained for each of the variants mentioned above, so that intermediate positions between the locations can be determined even more easily at defined times.

In a further step of the method, which is not illustrated in figures, the reception of the trajectory data and a calculation of the possible ego trajectories takes place. The ego vehicle receives the trajectory data of the equipped vehicles and calculates several possible ego trajectories for itself. Then a collision area is defined as the intersection of the trajectories. The probability of collision is the probability that the vehicle geometries will touch or even overlap. One way to test this is to pull samples from the state probability densities of the two vehicles representing the respective trajectories (eg, using Unscented Kalman Filter, Random, Randomized Unscented Kalman Filter, Localized Cumulative Distribution, etc.) and then to Check whether the vehicle geometries overlap or touch one another at these samples, or whether the safety distance is not necessarily uniform in all directions.

If one has only estimated the geometry for the neighboring vehicle, one can also draw samples from it. Then you divide the number of all samples with collision by the total number of samples. Another possibility is to integrate over the product of the probability masses.

In a third step, a selection and transmission of the adapted data of the ego trajectory takes place: for the selection of the ego trajectory different criteria can be defined, such as e.g. lowest collision probability, collision probability lower than a time-dependent threshold, collision probability = 0 for the longest period, etc. Depending on the modeling of the probability distributions of the vehicle states (for example in Gauss), a collision probability = 0 is not possible. The selected trajectory is forwarded to a vehicle actuator and simultaneously sent via wireless communication (e.g., via V2X communication) as a new ego trajectory.

The described voting procedure can also be applied to other scenarios in which cooperation between vehicles makes sense, e.g. Entry and exit scenarios, overtaking maneuvers, traffic diversion due to a traffic jam situation or to reduce the burden of exhaust emissions, etc.

4 shows in principle a situation in which the second vehicle 20 a range of residence probabilities p ₂₀ created. As a result, a calculation of a collision area KB as an intersection of the trajectories of the two vehicles 10 . 20 be performed. Shown is a situation at a time of t = 5 s. The collision probability then corresponds to the probability that both vehicles are in the collision area at time t.

Although the scenarios explained above only for a first and second vehicle 10 . 20 It goes without saying that the proposed method can also be used for a larger number of vehicles. The vehicles 10 . 20 can be fully automated, semi-automatic or manually controlled. Preferably, the trajectory data becomes mutual between the two vehicles 10 . 20 exchanged so that even better maneuver planning can be carried out. However, the method can also be carried out with a unidirectional transmission of trajectory data of a vehicle to another vehicle.

5 shows a block diagram of an embodiment of the proposed device 100 for preparing a maneuver planning of at least one vehicle 10 . 20 , One recognizes a detection device 110 for determining a trajectory of a first vehicle 10 wherein the trajectory is defined as a defined number of space-time location areas of the first vehicle 10 are created with an associated probability distribution, wherein for defined times each areas are determined per assigned probability. Functional with the detection device 110 is a transmission device 120 for transmitting the trajectory data by wireless vehicle communication to a second vehicle 20 intended. The above operations take place in the first vehicle 10 instead of.

6 shows a block diagram of an embodiment of the proposed device 100 for performing a maneuver planning of at least one vehicle 10 . 20 , One recognizes a receiving device 210 for receiving the trajectory data of the first vehicle 10 , Functional with the receiving device 210 is a planning device 220 for performing a maneuver planning of the second vehicle 20 intended.

7 shows a basic sequence of a method for preparing a maneuver planning of at least one vehicle 10 . 20 ,

In one step 300 will be determining trajectory data of a first vehicle 10 in the form of a defined number of spatiotemporal areas of residence of the first vehicle 10 each with associated probability distribution, wherein for defined times each areas with associated probability p ₁₀ be determined carried out.

In one step 310 transmitting the trajectory data by means of vehicle communication to at least one further vehicle 20 carried out.

Advantageously, the inventive methods can be implemented as software, for example, on the electronic devices 100 . 200 expire. The electronic devices 100 . 200 can be used as control units of the vehicles 10 . 20 be educated. A simple adaptability of the method is supported in this way.

As a result, the proposed methods can advantageously increase a safety level in traffic and provide a homogeneous traffic flow.

Although the proposed methods have been described above exclusively on the basis of legal traffic, it goes without saying that the proposed method can also be used for left-hand traffic, in which case the direction / orientation information mentioned above must be adapted accordingly.

One skilled in the art will suitably modify and / or combine the features of the invention without departing from the gist of the invention.

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• WO 17029096 A1 [0003]
• DE 102012011994 A1 [0004]
• WO 2017/076593 A1 [0005, 0006, 0027, 0033]
• DE 102016113903 A1 [0007]
• WO 2017/029096 A1 [0027]

Cited non-patent literature

• A. Bry and N. Roy, "Rapidly Exploring Random Trees for Motion Planning and Uncertainty", IEEE International Conference on Robotics and Automation (ICRA), 2011 [0035]

Claims (10)

Method for preparing a maneuver planning of at least one vehicle (10, 20), comprising the steps of: - determining trajectory data of a first vehicle (10) in the form of a defined number of space-time parameters of the first vehicle (10), each having assigned probability distribution for defined times respectively areas with assigned probability (p ₁₀ ) are determined; and - transmitting the trajectory data by means of vehicle communication to at least one further vehicle (20). Method according to Claim 1 wherein the trajectory data includes at least one of the following state histories of the first vehicle (10) over time: position, orientation, velocity, acceleration. Method according to Claim 1 or 2 , where the trajectory data are displayed according to a Gaussian mixing distribution in the longitudinal and transverse directions. Method according to one of Claims 1 to 3 wherein the trajectory data comprises a matrix of occupancy probabilities of cells into which the environment is divided. Method according to one of the preceding claims, wherein for determining the trajectory data, a defined collision probability of the first vehicle (10) with the at least one further vehicle (20) is taken into account. Method according to one of the preceding claims, wherein a mutual exchange of trajectory data between the first vehicle (10) and the at least one further vehicle (20) is performed. A method for performing a maneuver planning of a vehicle (10, 20), comprising the steps: - Receiving trajectory data of a first vehicle (10) in the form of a defined number of space-time parameters of the first vehicle (10), each with associated probability distribution, wherein defined points in time respectively areas with an associated probability (p ₁₀ ) are determined by a second vehicle (20); and - using the trajectory data of the first vehicle (10) to perform a maneuver planning of the second vehicle (20). Apparatus (100) for preparing maneuver planning of at least one vehicle (10, 20), comprising: - A determination device (110) for determining trajectory data of a first vehicle (10), wherein the trajectory data as a defined number of spatiotemporal residence areas of the first vehicle (10) are created with an associated probability distribution, wherein for defined times each areas per assigned Probability to be determined; and - A transmission device (120) for transmitting the trajectory data by vehicle communication to at least one further vehicle (20). Apparatus (200) for performing a maneuver planning of at least one vehicle (10, 20), comprising: - Receiving means (210) for receiving trajectory data of a trajectory of a first vehicle (10), wherein the trajectory as a defined number of spatiotemporal residence areas of the first vehicle (10) are created with an associated probability distribution, wherein for defined times each areas be determined per allocated probability; and a planning device (220) for planning the maneuver planning of the vehicle by means of the received trajectory data of the first vehicle. Computer program product with program code means for carrying out the method according to one of Claims 1 to 7 when it runs on an electronic device (100, 200) or is stored on a computer-readable medium.

Images