Trajectory Validation for Autonomous Driving

Abstract

The present invention relates to trajectory verification for autonomous driving. A method for determining whether a planned trajectory of a first vehicle on a road along which a first vehicle and a second vehicle are traveling is invalid, the method comprising: obtaining (S41) a planned trajectory (24), the planned trajectory comprising a first state of the first vehicle for each of a plurality of moments; obtaining (S42) a second state of the second vehicle for each of a plurality of moments; determining (S43) a corresponding lateral range extending from the second vehicle for each moment; and determining (S44) that the planned trajectory is invalid in the following circumstances: for the first state and the second state at one or more of the plurality of moments: the first vehicle is within the lateral range and within a lane boundary region; and the direction of the lateral velocity of the first vehicle is toward the second vehicle, and the lateral acceleration of the first vehicle away from the second vehicle is less than a predetermined threshold.

Description

Technical Field

[0001] Example aspects herein relate generally to the field of autonomous driving, and specifically to techniques for determining whether a planned trajectory used in autonomous control of a vehicle is invalid. Background Art

[0002] Autonomous driving functions are a feature of modern vehicles that have attracted increasing attention. Autonomous driving functions may allow the driver of a host vehicle (i.e., a vehicle being autonomously controlled) to hand over control of the acceleration and steering of the vehicle to an autonomous driving system that may be provided with a target speed and more detailed information of the forward time or intended route. The autonomous driving system may then attempt to achieve the desired speed by accelerating and steering the vehicle to follow the selected lane.

[0003] Autonomous driving systems can also be adapted to react appropriately to the actions of other road users. For example, when the host vehicle approaches a slower moving vehicle in front of it, the autonomous driving system can decide whether to overtake the slower moving vehicle or slow down and maintain a desired forward distance to the vehicle in front. Autonomous driving systems can additionally switch lanes to follow the desired route. More advanced versions of the system can even predict the behavior of other road users to determine appropriate actions and reactions. Therefore, autonomous driving systems are typically configured to obtain information from devices such as radars, cameras, inertial measurement units, etc., in order to collect data about the host vehicle and its environment, and generate a high-level environmental model that describes the road and the traffic on the road.

[0004] The autonomous driving system may then also be arranged to identify one or more maneuvers that the host vehicle can perform based on the generated high-level environment, to select a maneuver to perform and to determine how the maneuver should be performed (in other words, to determine the trajectory of the host vehicle and the appropriate control signals required to achieve the determined trajectory, such as the acceleration and steering angle of the host vehicle), and to control the host vehicle to perform the determined maneuver. The two main approaches to autonomous driving algorithms are rule-based, and statistical models including those based on machine learning, cost functions, etc. Summary of the invention

[0005] According to a first aspect of the present invention, the inventors have devised a computer-implemented method for determining whether a planned trajectory of a first model vehicle used in autonomous control of a vehicle modeled by a first model vehicle on a model of a road along which the first model vehicle and a second model vehicle are traveling is invalid, the model of the road comprising a plurality of lanes defined by a plurality of lane boundaries. The method comprises the following steps: obtaining a planned trajectory comprising a corresponding first vehicle state of the first model vehicle defined relative to the model of the road for each of a plurality of moments, each first vehicle state being defined by a corresponding lateral position of the first model vehicle, a corresponding lateral velocity of the first model vehicle, and a corresponding lateral acceleration of the first model vehicle. The method further comprises the following steps: obtaining a second vehicle state of the second model vehicle defined relative to the model of the road for each of the plurality of moments, each second vehicle state being defined by a corresponding lateral position of the second model vehicle in the model of the road. The method further comprises the following steps: determining a corresponding lateral range extending from the second model vehicle for each of a plurality of moments using a corresponding lateral position of the second model vehicle relative to the center of a lane in which the second model vehicle is located among a plurality of lanes. The method also includes the following steps: determining that the planned trajectory is invalid under the following circumstances: for the first vehicle state and the second vehicle state at at least one of the multiple moments, the following conditions in a first condition set are satisfied: (i) the lateral position of the first model vehicle is within the lateral range of the second model vehicle at the moment and is within a lane boundary area, and the lane boundary area extends along and includes a lane boundary among the multiple lane boundaries; and (ii) the direction of the lateral velocity of the first model vehicle is toward the second model vehicle, and the magnitude of the lateral acceleration of the first model vehicle away from the second model vehicle is less than a predetermined lateral deceleration of the first model vehicle.

[0006] According to the second aspect of the present invention, the inventors also design a computer program comprising instructions, which, when executed by a computer processor, cause the computer processor to perform the method according to the first aspect. The computer program may be stored on a non-transitory computer-readable storage medium or carried by a signal.

[0007] According to the third aspect of the present invention, the inventors have also designed a device, which is arranged to determine whether a planned trajectory of the first model vehicle used in the autonomous control of a vehicle modeled by a first model vehicle on a model of a road along which the first model vehicle and a second model vehicle are traveling is invalid, and the model of the road includes a plurality of lanes defined by a plurality of lane boundaries. The device includes a planned trajectory acquisition module, which is configured to obtain a planned trajectory, which includes a corresponding first vehicle state of the first model vehicle defined relative to the model of the road for each of a plurality of moments, each first vehicle state being defined by a corresponding lateral position of the first model vehicle, a corresponding lateral velocity of the first model vehicle, and a corresponding lateral acceleration of the first model vehicle. The device also includes a second vehicle state acquisition module, which is configured to obtain a second vehicle state of the second model vehicle defined relative to the model of the road for each of the plurality of moments, each second vehicle state being defined by a corresponding lateral position of the second model vehicle in the model of the road. The device also includes a lateral range determination module, which is configured to use the corresponding lateral position of the second model vehicle relative to the center of the lane in which the second model vehicle is located in the plurality of lanes to determine the corresponding lateral range extending from the second model vehicle for each of the plurality of moments. The device also includes a validity determination module, which is configured to determine that the planned trajectory is invalid in the following case: for the first vehicle state and the second vehicle state at at least one of the plurality of moments, the following conditions in the first condition set are satisfied: (i) the lateral position of the first model vehicle is within the lateral range of the second model vehicle at the moment and is within a lane boundary area, the lane boundary area extends along and includes one of the plurality of lane boundaries; and (ii) the direction of the lateral velocity of the first model vehicle is toward the second model vehicle, and the magnitude of the lateral acceleration of the first model vehicle away from the second model vehicle is less than a predetermined lateral deceleration of the first model vehicle.

[0008] According to a fourth aspect of the present invention, the inventors also design a vehicle, the vehicle comprising an automatic driving system and a device according to the first aspect, the device being arranged to determine whether a planned trajectory is invalid. The device is further arranged to determine that the planned trajectory is a valid planned trajectory if the planned trajectory is not determined to be invalid, and output the valid planned trajectory to the automatic driving system, the automatic driving system being arranged to autonomously control the vehicle to drive along a road according to the valid planned trajectory. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Embodiments of the present disclosure will now be explained in detail, by way of non-limiting example only, with reference to the accompanying drawings described below. The same reference numerals appearing in different figures of the accompanying drawings may represent identical or functionally similar elements, unless otherwise specified.

[0010] Figure 1A is a schematic illustration of an apparatus arranged to determine whether a planned trajectory of a first model vehicle used in autonomous control of a vehicle modeled by the first model vehicle is invalid according to an example embodiment herein.

[0011] Figure 1B is a schematic illustration of a vehicle according to an example implementation herein.

[0012] Figure 2A An example of a model of a road defined in a lane coordinate system is illustrated.

[0013] Figure 2B Illustrated Figure 2A Example of a planned trajectory for a first model vehicle on a model of the road.

[0014] Figure 3 is a block diagram illustrating an example implementation of an apparatus of an example embodiment in programmable signal processing hardware.

[0015] Figure 4 is a flow chart illustrating a process of determining a target lateral acceleration of a host vehicle by an apparatus according to an example embodiment.

[0016] Figure 5 is a graph illustrating the change in left range calculated using the lateral range function described herein.

[0017] Fig. 6A is a schematic illustration showing an example of a lateral range determined by a lateral range determination module of the apparatus of FIG. 1 .

[0018] Figure 6B is a schematic illustration showing an example of a lateral range determined by the lateral range determination module of the apparatus 10 of FIG. 1 at a specific moment when the second model vehicle moves into the lane boundary area.

[0019] Figure 6C The lateral range determination module determines Fig. 6A Schematic illustration of a first example of the left and right parts of the lateral extent.

[0020] Fig.6D The lateral range determination module determines Fig. 6A Schematic illustration of a second example of the left and right parts of the lateral extent of.

[0021] Figure 7is a schematic illustration showing an example of a braking area determined by the apparatus of FIG. 1 .

[0022] Figure 8 is a diagram showing an exemplary embodiment of the present invention. Figure 4 Schematic illustration of an example of all regions defined by the process.

[0023] Fig.9A and Fig. 9B is a diagram showing that for each of the first unit O_1 and the second unit O_n in the plurality of units representing the second model vehicle O, Figure 4 Schematic illustration of an example of all regions defined by the process. DETAILED DESCRIPTION

[0024] In order to be considered roadworthy, vehicles incorporating autonomous driving capabilities are typically required to provide a high level of safety. For example, while the reliance on sensing devices for safety can be demonstrated only with respect to likelihood, the algorithms used by autonomous driving systems for behavior control and trajectory planning can be demonstrated to be safe using formal models such as Responsibility Sensitive Safety (RSS). RSS is a formal model that describes permissible behaviors that ensure safety.

[0025] However, in implementing an autonomous driving system such that the determined behavior and trajectories of an autonomously controlled vehicle are safe and similar to human behavior (i.e., such that autonomous control of the vehicle corresponds to manual control performed by a safe human driver), and in formally proving the safety of the autonomous driving system, various difficulties may arise that must be overcome in order to provide autonomous driving that is suitable for driving.

[0026] As an example, autonomous driving algorithms based on statistical models may lack transparency compared to rule-based models. Rule-based models allow predictions of what will happen in any given situation, and link decisions about subsequent actions to be taken to specific rules. In contrast, collisions caused by decisions made by statistical models (such as, machine learning models) may be more difficult to prove, and the causes may be more difficult or impossible to determine. The inventors have recognized that when generalized to any number of vehicles in a host vehicle environment, the behavior in many conventional autonomous driving algorithms based on cost functions may have to be analyzed on a case-by-case basis with respect to the number of vehicles and all possible constellations of these vehicles in order to determine safety. Importantly, the trajectories generated by any of these algorithms can be verified and therefore considered safe before being used to control the host vehicle.

[0027] As such, other advantageous autonomous driving algorithms based on statistical models such as machine learning or cost functions with optimizers may be difficult to analyze to the extent that mathematical safety guarantees can be made.

[0028] Therefore, rather than analyzing autonomous algorithms directly based on statistical models, the inventors have recognized that simpler interpretable safety models can be developed, which can then be applied to trajectories generated by autonomous driving algorithms and used to validate these trajectories after they are generated.

[0029] As an example, a trajectory generated by an autonomous algorithm based on a statistical model may be evaluated using the rules of an RSS model that defines a safe distance to be maintained between vehicles. However, while the RSS model is sufficient to demonstrate safety with respect to emergency maneuvers, as the model ensures that no vehicle will collide from a safe state, it does not define conditions for safe behavior with respect to road lanes. Thus, the simplest strategy that complies with the rules of the RSS model may result in a situation that allows a vehicle to be stuck between other vehicles in a lane, for example, if the vehicle has avoided an immediate danger but cannot return to a safe lateral distance.

[0030] Furthermore, the distances defined by the RSS model are based on anticipated emergency maneuvers that, although safe, may not guarantee comfortable, human-like driving.

[0031] Oskar Larsson's master's thesis "The Oskillator, Artificial Force Field Highway Chauffeur" (Chalmers University of Technology, Gothenburg, Sweden 2019) (https: / / hdl.handle.net / 20.500.12380 / 300733) proposed an improved responsibility-sensitive safety (MRSS) model, the entire content of which is incorporated herein. In the MRSS model, the safe lateral distance is defined as the minimum lateral avoidance distance, but not only the minimum lateral avoidance distance to the other vehicle itself, but also the minimum lateral avoidance distance to all lanes within the minimum lateral avoidance distance to other vehicles. This method allows to ensure that collisions between vehicles are avoided, and also avoids vehicles turning into the same lane, so unsafe situations may occur. In this way, the method takes into account the required behavior of the host vehicle relative to the road lanes.

[0032] However, the present inventors have recognised that none of these safety models take into account how the host vehicle should respond when other road users engage in unsafe driving that violates the rules defined by these safety models.

[0033] As an example, unsafe driving may include, for example, a situation where another road user following the host vehicle is dangerously close to the host vehicle, i.e., following the host vehicle at a distance less than the emergency distance at which the following road user would need to perform an emergency braking maneuver. Another example of unsafe driving may occur when another road user moves into the lane in which the host vehicle is traveling at a dangerously close distance behind or in front of the host vehicle.

[0034] In this case, for example, a human driver can control their vehicle to stay in the same lane or switch to another lane. Such a maneuver would be considered by a safe human driver to represent the best response to a dangerous situation caused by other road users. However, for both maneuvers, the application of the rules of the RSS model or the MRSS model would render the corresponding trajectory generated by the autonomous driving algorithm invalid, because according to the rules of these models, the host vehicle would initially be dangerously close to other road users.

[0035] As such, the inventors have recognised that application of the rules of existing safety models may result in trajectories generated by an autonomous driving algorithm being invalidated as dangerous, even in circumstances where such trajectories represent an optimal response to a dangerous situation caused by unsafe driving by another road user (i.e., any safety issue is unavoidable and does not immediately endanger the driver or other road users).

[0036] Furthermore, the inventors have recognized that some existing safety models, such as the MRSS model, do not discuss how to use defined rules to verify the actual trajectories generated by the autonomous driving algorithm.

[0037] The example embodiments described below may address one or more of the problems outlined above, and will now be described in detail with reference to the accompanying drawings. Where a reference numeral is attached to a technical feature in the drawings, detailed description, or any claim, the reference numeral is included solely for the purpose of increasing the intelligibility of the drawings, detailed description, and claims. Therefore, the reference numeral and its absence have no limiting effect on the scope of any claim element.

[0038] Figure 1A is a schematic illustration of an apparatus 10 arranged to determine whether a planned trajectory of a first model vehicle used in autonomous control of a vehicle 1 modelled by the first model vehicle is invalid according to an example embodiment herein.

[0039] More specifically, the apparatus 10 is arranged to determine that the first model vehicle is on the road 20 (at Figure 2A Is the planned trajectory on the model shown in (shown in) invalid, the first model vehicle and the second model vehicle O (in Figure 2A) along a road 20, the planned trajectory being used in autonomous control of a vehicle 1 modeled by a first model vehicle. The model of the road 20 includes a plurality of lanes 21A, 21B, 21C defined by a plurality of lane boundaries. As an example, as in this example embodiment, the vehicle 1 may be considered a host vehicle (i.e., a vehicle being autonomously controlled).

[0040] Figure 2A An example of a model of a road 20 defined in a lane coordinate system is illustrated. The model of the road 20 is an example of a modeled road along which the first model vehicle and the second model vehicle O can travel.

[0041] exist Figure 2A In the example of FIG. 2 , a road 20 has a plurality of lane boundaries 22A, 22B, 22C, 22D (in Figure 2B 21A, 21B, and 21C are defined as shown in FIG. 21A . As in this example, the lanes of the road 20 may have the same width. As in this example, one of the lanes 21A, 21B, and 21C may be used as a so-called “fast lane,” and another of the lanes 21A, 21B, and 21C may be used as a so-called “slow lane.” In this example, the leftmost lane 21A serves as a fast lane, and the rightmost lane 21C serves as a slow lane.

[0042] Figure 2A The exemplary model of the road 20 shown illustrates the road 20 and traffic on the road at a particular time. In the exemplary model of the road 20, a first model vehicle is located at a position 26 in a middle lane 21B and is traveling in a direction indicated by an arrow 27. A second model vehicle O (e.g., a vehicle of another road user, i.e., a road user different from the first model vehicle) is traveling in the rightmost lane 21C in the same direction as the first model vehicle.

[0043] As in this example, the positions of the first model vehicle and the second model vehicle O in the model of the road 20 can be defined relative to the center of the bounding box of each vehicle. As an alternative, the positions of the first model vehicle and the second model vehicle O can be defined relative to any other predetermined reference point on the model vehicle (e.g., a predetermined corner of the bounding box of the model vehicle, the center of gravity of the model vehicle, the center of mass of the model vehicle, etc.).

[0044] although Figure 2A The model of road 20 shown in FIG. 1 has three lanes, but the road on which the model of road 20 is based may have one lane, two lanes, or four or more lanes, and a corresponding number of lanes may be included in the model of road 20. As another alternative, although in FIG. Figure 2AA single second model vehicle O is shown in the model of the road 20 shown, but any suitable number of model vehicles in addition to the first model vehicle may be included in the model of the road 20, and optionally the number of additional model vehicles may depend on the traffic on the road at a given time. As an example, the number of additional model vehicles may be determined based on information from devices such as radars, cameras, inertial measurement units, etc., which collect data about the vehicle 1 and its environment in order to generate a high-level environmental model describing the road and the traffic on the road, i.e., the model of the road 20.

[0045] exist Figure 2A In the example of FIG. 1 , the model of the road 20 is defined in a lane coordinate system. The lane coordinate system may be a two-dimensional curvilinear coordinate system suitable for reflecting the road on which the model of the road 20 is based. In particular, the curvilinear coordinate system constituting the lane coordinate system has two axes, namely, a longitudinal axis or an x-axis extending in a longitudinal direction along the road (in Figure 2A ) and a transverse axis or y-axis extending across the road (indicated by the reference numeral x in Figure 2A The x-axis is always parallel to the lanes of the road, and the y-axis is orthogonal to the x-axis at each value of x. Alternatively, the lane coordinate system, and therefore the model of the road 20, may be defined by any other suitable means, such as by using a Cartesian coordinate system or any other suitable coordinate system.

[0046] More generally, the lane coordinate system and the model of the road 20 can be defined in any suitable manner so that the device 10 is provided with information about the position of the first model vehicle, the number and width of lanes 21A, 21B, 21C, the position and speed of the second model vehicle O and any additional model vehicles on the road, and optionally also the curvature of the road 20.

[0047] exist Figure 2A In the example lane coordinate system shown, the x-axis of the lane coordinate system increases in the forward travel direction of the first model vehicle, and the y-axis of the lane coordinate system increases in the direction of the leftmost lane 21A of the model of the road 20. However, the lane coordinate system may be oriented in any other suitable manner. As an example, the x-axis of the lane coordinate system may increase in a direction opposite to the forward travel direction of the first model vehicle, and / or the y-axis may increase in the direction of the rightmost lane of the model of the road 20.

[0048] In addition, as in Figure 2A In the example of , the lane coordinate system may be defined such that x=0 represents the position of the first model vehicle on the x-axis, and y=0 represents the center of the lane of the road 20. In particular, as in Figure 2AIn the example of , the lane coordinate system may be defined such that y=0 represents the center of the center lane 21B. However, the lane coordinate system may be defined in any other suitable manner, for example, y=0 may be selected to represent the center of any lane in the multiple lanes of the model of the road 20 .

[0049] like Figure 1A As shown, the device 10 includes a planning trajectory obtaining module 11 , a second vehicle state obtaining module 12 , a lateral range determining module 13 and a validity determining module 14 .

[0050] The planned trajectory acquisition module 11 is configured to obtain a planned trajectory, which includes a corresponding first vehicle state defined by the model of the first model vehicle relative to the road for each of multiple moments, and each first vehicle state is defined by a corresponding lateral position of the first model vehicle, a corresponding lateral speed of the first model vehicle, and a corresponding lateral acceleration of the first model vehicle.

[0051] In particular, it is assumed that sufficient information and data about the vehicle 1 and its environment is available from devices such as radars, cameras, inertial measurement units, etc., so that the model of the road 20 generated thereby can describe the estimated future state of the vehicle's environment (e.g., the road and the traffic on the road). Figure 2A The example of the model of road 20 shown illustrates the road and the traffic on the road, i.e., the estimated state of the environment, at a specific time, but as in the present example embodiment, the model of road 20 may also describe the estimated state of the environment at multiple future time points. For example, the model of road 20 may describe the estimated state of the environment at different time points during a set time period in the future starting from, for example, the current time point or the next future time point.

[0052] As a more specific example, the set period may be between 0.1 seconds and 5 seconds, and the model of the road 20 may include an estimated state of the environment for each of a plurality of moments in the set period. The plurality of moments may correspond to a plurality of time steps with a set time interval between each pair of consecutive moments. For example, the model of the road 20 may include a plurality of estimated states of the environment at 0.25 second intervals within a 5 second period. The interval should be short enough to use linear interpolation or an assumption of constant velocity or acceleration to produce a sufficiently accurate representation of the vehicle state at that time.

[0053] Figure 2B Illustrated Figure 2A 20. Specifically, Figure 2B A segment of a model of a road 20 is illustrated over which the road is approximately linear.

[0054] In the model of road 20, as in Figure 2BIn the present example shown in , the y-axis may be scaled to the lane width, such that each increment or decrement of 1 on the y-axis represents the width of one lane, such that y=1 represents the center of the leftmost lane 21A, and y=-1 represents the center of the rightmost lane 21C, and y={-1.5, -0.5, 0.5, 1.5} represents lane boundaries 22A, 22B, 22C, 22D of lanes 21A, 21B, and 21C. Alternatively, the y-axis may be scaled in any other suitable manner, such as such that each increment or decrement of 1 on the y-axis represents half the width of one lane, such that lane centers and lane boundaries have integer values, or the y-axis may not be scaled.

[0055] Additionally or alternatively, such as Figure 2B In the example shown, the model of the road 20 may include a plurality of lane boundary regions 23A, 23B, 23C, 23D. Each lane boundary region 23A, 23B, 23C, 23D extends along and includes a corresponding lane boundary 22A, 22B, 22C, 22D of the plurality of lane boundaries.

[0056] As in this example, each lane boundary region may use a lateral offset Δy from the center of the corresponding lane bias To define the lateral offset Δy bias The value of can be selected to represent the acceptable deviation (bias leeway) from the center of the lane that the vehicle may have while still being considered to be completely (safely) within the lane. In particular, to ensure safety, unless the first model vehicle is switching lanes, the first model vehicle should stay near the center of the lane it is traveling in, with a lateral offset of Δy bias may be selected to represent the maximum lateral deviation from the center of the lane that the first model vehicle may appropriately have when not switching lanes. Thus, the lane boundary area may correspond to a portion on either side of the lane boundary 22A, 22B, 22C, 22D of the respective lane 21A, 21B, 21C that is greater than the lateral offset Δy from the center of the lane 21A, 21B, 21C. bias .

[0057] As a more specific example, in this example, the lane boundary area R l It can be defined as follows:

[0058] (1)

[0059] in,

[0060] The planned trajectory 24 includes a first vehicle state of the first model vehicle relative to the model definition of the road 20 for each of the plurality of time instants. Figure 2B In the example shown, the planned trajectory includes nine first vehicle states defined by respective lateral positions 25_1 to 25_9 of the first model vehicle. Figure 2B Not shown in FIG. 1 , but each of the first vehicle states is also defined by a respective lateral velocity of the first model vehicle and a respective lateral acceleration of the first model vehicle.

[0061] So, in Figure 2B In the example shown, the planned trajectory 24 defines the maneuver by defining the state of the first model vehicle in terms of lateral position, lateral velocity, and lateral acceleration (i.e., first vehicle state) at each of nine consecutive moments in time while executing the maneuver in nine steps of moving the first model vehicle from lane 21B to lane 21A. In this example, the plurality of moments includes nine moments in time. However, as described above, any suitable number of moments may be considered.

[0062] Optionally, the corresponding first vehicle state at each of the plurality of moments may also include any additional information defining the state of the first model vehicle. As a non-limiting example, the first vehicle state may additionally include one or more of a longitudinal position of the first model vehicle, a longitudinal velocity of the first model vehicle, a longitudinal acceleration of the first model vehicle, a lateral extent or width of the first model vehicle, and a longitudinal extent or length of the first model vehicle.

[0063] As an example, in this example implementation, the first vehicle state at each of the multiple time points may be defined by the following information of the first model vehicle:

[0064] l- longitudinal extent (length)

[0065] w - lateral extent (width)

[0066] y-horizontal position

[0067] ·v x - Longitudinal speed

[0068] ·v y - Lateral speed

[0069] ·a x - Longitudinal acceleration

[0070] ·a y - Lateral acceleration.

[0071] More generally, as in this example, the first model vehicle and any additional model vehicles may be represented as objects having dynamic characteristics defined in a lane coordinate system as described above. As will be discussed in more detail below, the first model vehicle and any additional model vehicles may alternatively be represented as one or more cells of a grid having dynamic characteristics defined in a lane coordinate system.

[0072] Reference again Figure 1A The second vehicle state acquisition module 12 is configured to obtain the second vehicle state defined by the second model vehicle O relative to the model of the road 20 for each of a plurality of moments, each second vehicle state being defined by a corresponding lateral position of the second model vehicle O in the model of the road 20.

[0073] As an example, similar to the first vehicle state, as in this example, the second vehicle state for each of the plurality of moments in time can be represented as an object having dynamic characteristics defined in a lane coordinate system. As will be discussed in more detail below, the first model vehicle and any additional model vehicles can alternatively be represented as one or more cells of a grid having dynamic characteristics defined in a lane coordinate system.

[0074] As a more specific example, as in the present exemplary embodiment, the second model vehicle O may be defined by coordinates identifying the position of the second model vehicle O in the lane coordinate system of the model of the road 20, the length of the second model vehicle along the longitudinal axis of the lane coordinate system, and the width of the second model vehicle along the transverse axis of the lane coordinate system. Alternatively, the second model vehicle O may be defined by one or more cells of a grid defined in the lane coordinate system, as will be discussed in more detail below.

[0075] As another example, the corresponding second vehicle state at each of the multiple moments may optionally also include any additional information that defines the state of the second model vehicle O. As a non-limiting example, the second vehicle state may further include one or more of the longitudinal position of the second model vehicle O, the longitudinal speed of the second model vehicle O, the lateral speed of the second model vehicle O, the longitudinal acceleration of the second model vehicle O, the lateral acceleration of the second model vehicle O, the longitudinal extent or length of the second model vehicle, and the lateral extent or width of the second model vehicle O.

[0076] As an example, as in the present example embodiment, the second vehicle state at each of the plurality of time points may be defined by the following information of the second model vehicle O (wherein the subscript i indicates that the second model vehicle O may be the i-th model vehicle among one or more model vehicles defined in the model of the road 20 in addition to the first model vehicle):

[0077] ·l i - Longitudinal extent (length)

[0078] ·w i - Horizontal extent (width)

[0079] ·x i -Vertical position

[0080] ·y i - Horizontal position

[0081] ·v x,i - Longitudinal speed

[0082] ·v y,i - Lateral speed

[0083] ·a x,i - Longitudinal acceleration

[0084] ·a y,i - Lateral acceleration.

[0085] The lateral range determination module 13 is configured to determine a respective lateral range extending from the second model vehicle O for each of a plurality of times using the respective lateral positions of the second model vehicle O relative to the center of the lane 21C in which the second model vehicle O is located.

[0086] The validity determination module 14 is configured to determine that the planned trajectory is invalid when: for the first vehicle state and the second vehicle state at at least one of the plurality of time instants, the following conditions in the first condition set are satisfied:

[0087] (i) the lateral position of the first model vehicle is within the lateral range of the second model vehicle O at this moment and is within a lane boundary area 23A, 23B, 23C, 23D extending along and including lane boundaries 22A, 22B, 22C, 22D among the plurality of lane boundaries; and

[0088] (ii) The direction of the lateral velocity of the first model vehicle is toward the second model vehicle O, and the magnitude of the lateral acceleration of the first model vehicle away from the second model vehicle O is smaller than the predetermined lateral deceleration of the first model vehicle.

[0089] Figure 1B is a schematic illustration of a vehicle 1 according to an example embodiment herein. As in this example embodiment, vehicle 1 may be considered a host vehicle (ie, an autonomously controlled vehicle) and serves as an example of a vehicle that may be modeled by a first model vehicle.

[0090] The vehicle 1 comprises an automated driving system 15 and an apparatus 10 arranged to determine whether a planned trajectory is invalid. Figure 1B In an example embodiment of , the apparatus is further arranged to determine that the planned trajectory is a valid planned trajectory and output the valid planned trajectory to the autonomous driving system 15 if the planned trajectory is not determined to be invalid.

[0091] The autonomous driving system 15 is arranged to autonomously control the vehicle 1 to drive along the road according to the valid planned trajectory. As an example, the autonomous driving system 15 may be arranged to generate control signals for controlling the steering and / or acceleration of the vehicle 1 so that the vehicle 1 performs maneuvers according to the valid planned trajectory, or to control the vehicle 1 by any other suitable means.

[0092] The combination of the automatic driving system 15 and the device 10 may be referred to as an autonomous driving system, i.e., an autonomous driving system capable of performing behavior and trajectory planning and subsequent control of the host vehicle.

[0093] exist Figure 1B In the example embodiment shown, the apparatus 10 and the autonomous driving system 15 are illustrated as separate devices. The functionality of the apparatus 10 and the autonomous driving system 15 may alternatively be provided by a suitably configured single device (e.g., a suitably programmed computer processor).

[0094] Alternatively, for example, the apparatus 10 can be used to determine whether a trajectory is invalid as part of a simulation of an autonomously controlled vehicle, or as part of testing and analysis of an autonomous driving algorithm (such as an autonomous driving algorithm based on a statistical model). For example, when training an artificial intelligence or machine learning model, it is often useful to have a termination condition, such as in reinforcement learning, for example where the agent learns successfully or fails. In this way, a valid planned trajectory suitable for use in autonomous control of a vehicle 1 modeled by a first model vehicle may not necessarily be used to control the vehicle 1.

[0095] As is apparent from this disclosure, in order for the validity determination module 14 of the apparatus 10 to determine that the planned trajectory is invalid, autonomously controlling the vehicle 1 according to the planned trajectory must at least at one time bring the first model vehicle within a certain lateral distance of the second model vehicle O, i.e., within the lateral range of the second model vehicle. In this way, if the safe lateral distance to the second model vehicle is not maintained, then Figure 1A The apparatus 10 may allow a planned trajectory to be determined to be invalid.

[0096] However, in order for the validity determination module 14 to determine that the planned trajectory is invalid, at least at one time when the lateral position of the first model vehicle is within the first lateral range, the lateral position of the first model vehicle must also be within the lane boundary area. In this way, when the lateral position of the first model vehicle at a specific time is within the area of ​​the lane 21C where the second model vehicle O is traveling, directly in front of the second model vehicle O, or within the area of ​​the lane 21C where the second model vehicle O is traveling, directly behind the second model vehicle O (i.e., when the lateral position of the first model vehicle at a specific time is within a certain distance from the center of the lane 21C where the second model vehicle is traveling), the above condition (i) is not satisfied.

[0097] Thus, in the case where the second model vehicle O represents another road user that is following the first model vehicle at a dangerously close distance or moving into the lane in which the first model vehicle is traveling at a dangerously close distance behind the first model vehicle, a planned trajectory that causes the vehicle 1 to be autonomously controlled to remain in the lane in which it is traveling will not be determined to be invalid, even though the lateral position of the first model vehicle will be within the determined lateral range of the second model vehicle at most, if not all, of the multiple moments for such a planned trajectory. In particular, for a planned trajectory that does not result in a lane switch, the first model vehicle will typically not enter the lane boundary region of the lane in which it is traveling, but will remain near the center of the lane, so that the second part of condition (i) is not satisfied.

[0098] Thus, the apparatus 10 may avoid determining as invalid a planned trajectory that represents an optimal response to another road user's unsafe driving (eg, staying in the same lane).

[0099] In addition, in order for the validity determination module 14 to determine that the planned trajectory is invalid, at least at a moment when the first model vehicle is within the lateral range of the second model vehicle O and within the lane boundary areas 23A, 23B, 23C, and 23D, the direction of the lateral velocity of the first model vehicle must be toward the second model vehicle O, and the amplitude of the lateral acceleration of the first model vehicle away from the second model vehicle O must be less than the predetermined lateral deceleration of the first model vehicle.

[0100] In the case where the second model vehicle represents another road user that is following the first model vehicle at a dangerously close distance or moves into a lane in which the first model vehicle is traveling at a dangerously close distance behind the first model vehicle, the planned trajectory that will cause the vehicle 1 to be autonomously controlled to switch to another lane will result in a lateral velocity of the first model vehicle being directed away from the second model vehicle O and / or a lateral acceleration of the first model vehicle being away from the second model vehicle O that is greater than or equal to a predetermined lateral deceleration of the first model vehicle. In particular, in the case where the second model vehicle O is driving in the same lane as the first model vehicle, the movement of the first model vehicle to the other lane results in an increase in the lateral distance between the first model vehicle and the second model vehicle O, which in turn requires the first model vehicle to accelerate laterally away from the second model vehicle O and / or the lateral velocity of the first model vehicle to move away from the second model vehicle O.

[0101] Therefore, in the case where the second model vehicle O represents another road user that is following the first model vehicle at a dangerously close distance or moving into the lane in which the first model vehicle is travelling at a dangerously close distance behind the first model vehicle, the planned trajectory that results in vehicle 1 being autonomously controlled to switch to a different lane will not be determined to be invalid because condition (ii) will not be satisfied.

[0102] Thus, the apparatus 10 may also avoid determining as invalid other planned trajectories that represent an optimal response to unsafe driving by another road user (eg, switching to another lane).

[0103] However, the device 10 may still determine an invalid planned trajectory that, for example, will result in the vehicle 1 being autonomously controlled to move to a dangerously close distance in the same lane as the second model vehicle O, or otherwise move to a dangerously close lateral distance from the second model vehicle O (i.e., resulting in a planned trajectory representing unsafe driving of the first model vehicle), because in these cases the first model vehicle will have to pass through a lane boundary area, the lateral velocity of the first model vehicle will need to be toward the second model vehicle O, and the first model vehicle will accelerate toward the second model vehicle O or have a minimum acceleration away from the second model vehicle O.

[0104] Thus, the device 10 can ensure that a planned trajectory that would cause the vehicle 1 to be autonomously controlled to drive unsafely can be invalidated.

[0105] In addition, by obtaining the first vehicle state and the second vehicle state for each of multiple moments, and determining whether the planned trajectory is invalid when conditions (i) and (ii) are satisfied for even one of the multiple moments, the device 10 can allow the safety of the entire planned trajectory to be analyzed.

[0106] As such, the apparatus 10 may solve one or more of the problems outlined above with respect to conventional approaches to trajectory verification.

[0107] Figure 3 3 is a schematic illustration of a programmable signal processing device 300 that can be configured to implement the functionality of the device 10. The signal processing device 300 includes an interface module 310 for receiving information and data about the vehicle 1 and its environment. The signal processing device 300 also includes a processor (CPU) 320 for controlling the device 10, a working memory 330 (e.g., a random access memory), and an instruction storage unit 340 storing a computer program, the computer program including computer readable instructions that, when executed by the processor 320, cause the processor 320 to perform processing operations of the device 10. The instruction storage unit 340 may include a ROM (e.g., in the form of an electrically erasable programmable read-only memory (EEPROM) or flash memory) preloaded with computer readable instructions. Alternatively, the instruction storage unit 340 may include a RAM or similar type of memory, and the computer readable instructions may be input therein from a computer program product (e.g., a computer readable storage medium 350, such as a CD-ROM, etc.) or a computer readable signal 360 carrying computer readable instructions.

[0108] In this example embodiment, the system includes a processor 320, a working memory 330, and an instruction storage unit 340. Figure 3 The illustrated combination of hardware components 370 is configured to implement Figure 1A The functionality of each of the component modules of device 10 is shown.

[0109] Figure 4 It is an example Figure 1A Flow chart of a process of determining whether a planned trajectory of a first model vehicle on a model of a road 20 is invalid, along which the first model vehicle and a second model vehicle O are traveling, the planned trajectory being used in autonomous control of a vehicle 1 modeled by the first model vehicle. The model of the road 20 includes a plurality of lanes defined by a plurality of lane boundaries.

[0110] exist Figure 4 In the processing step S41, the planned trajectory obtaining module 11 obtains the planned trajectory. The planned trajectory includes a corresponding first vehicle state of the first model vehicle defined relative to the model of the road 20 for each of the plurality of moments, each first vehicle state being defined by a corresponding lateral position of the first model vehicle, a corresponding lateral velocity of the first model vehicle, and a corresponding lateral acceleration of the first model vehicle.

[0111] The planned trajectory obtaining module 11 may be configured to obtain the planned trajectory in any suitable form.

[0112] For example, as in the present example embodiment, the device 10 may be configured to store a model of the road 20. As in the present example embodiment, the planning trajectory acquisition module 11 may be configured to receive the generated model of the road 20 by any suitable means, the model of the road including the estimated state of the environment of the vehicle at each of the multiple moments. As an alternative, in some example embodiments, the planning trajectory acquisition module 11 may be configured to receive information from a device (such as a radar, a camera, an inertial measurement unit, etc.) that collects data about the vehicle 1 and its environment. The information may be in the form of raw or processed data. In such an example embodiment, the planning trajectory acquisition module 11 may be configured to use the received information to generate a model of the road 20, the model of the road 20 including the estimated state of the environment of the vehicle at each of the multiple moments.

[0113] In this example embodiment where the device 10 stores a model of the road 20, as in the present exemplary embodiment, the planned trajectory acquisition module 11 can be configured to obtain the planned trajectory by receiving information describing the planned trajectory in any suitable form from any entity capable of generating a planned trajectory (e.g., a device implementing an autonomous driving algorithm based on a statistical model) and using the received information describing the planned trajectory and the stored model of the road 20 to determine the corresponding first vehicle state of the first model vehicle for each of a plurality of moments in time.

[0114] As an alternative, the planned trajectory acquisition module 11 can be configured to obtain the planned trajectory by receiving the corresponding first vehicle state of the first model vehicle for each of a plurality of moments (e.g., in the form of an ordered plurality of first vehicle states) from any suitable entity that can generate the first vehicle state. For example, a device implementing an autonomous driving algorithm based on a statistical model can also generate a trajectory, can determine a first vehicle state corresponding to the trajectory, and provide the first vehicle state to the planned trajectory acquisition module as the planned trajectory. In a simple form, the planned trajectory acquisition module 11 can, for example, be configured to obtain a set of values ​​for each of a plurality of moments as the corresponding first vehicle state of the first model vehicle defined by the model relative to the road 20 for each of a plurality of moments.

[0115] In such an example embodiment, the device 10 may also store a model of the road 20 or may simply be provided with a computer program that executes the model. Figure 4 The information required for the process of the road 20, such as the position of the center of the lane 21A, 21B, 21C of the road 20, the width of the lane 21A, 21B, 21C or the position of the lane boundary 22A, 22B, 22C, 22D, and the width or lateral offset Δy of the lane boundary area 23A, 23B, 23C, 23D bias Width.

[0116] As another example, the planning trajectory obtaining module 11 may be configured to obtain the planning trajectory by any suitable means known to those skilled in the art. For example, the planning trajectory obtaining module 11 may obtain the planning trajectory via a direct communication link (which may be provided by any suitable wired or wireless connection, such as a universal serial bus (USB) or Bluetooth). TM The planned trajectory is received by the planning trajectory acquisition module 11 through a communication link (connection) or an indirect communication link (which can be provided by a network including a local area network (LAN), a wide area network (WAN) and / or the Internet). In addition, the planned trajectory can be obtained by obtaining (for example, by reading from a storage medium such as a CD or a hard disk, or receiving via a network such as the Internet) a pre-prepared planned trajectory.

[0117] exist Figure 4 In the processing step S42, the second vehicle state acquisition module 12 obtains the second vehicle state defined by the second model vehicle O relative to the model of the road 20 at each of multiple moments, and each second vehicle state is defined by the corresponding lateral position of the second model vehicle O in the model of the road 20.

[0118] The second vehicle state obtaining module 12 can be configured in any form and through the above Figure 4 Any means discussed in processing step S41 of receiving the second vehicle state for each of the plurality of time instants.

[0119] exist Figure 4 In the processing step S43 of FIG. 1 , the lateral range determination module 13 uses the corresponding lateral position of the second model vehicle O relative to the center of the lane 21C in which the second model vehicle O is located among the multiple lanes to determine the corresponding lateral range 40 (at each of the multiple time instants) extending from the second model vehicle O. Fig. 6A ).

[0120] The respective lateral ranges 40 determined for each of the plurality of moments in time may correspond to safe lateral distances, e.g., the distance required to perform a lateral emergency maneuver that reduces the lateral speed until the vehicle 1 is traveling in a straight line. As such, the lateral ranges 40 may be considered to define a range on either side of the second model vehicle within which the presence of the second model vehicle would be expected to affect the autonomous control of the vehicle 1, e.g., by placing some limits on its lateral movement, speed, or acceleration, in order to ensure safe driving of the autonomously controlled vehicle 1.

[0121] As an example, as in the present example embodiment, the lateral range determination module 13 may use the corresponding lateral position of the second model vehicle O relative to the center of the lane 21C, and determine the corresponding lateral range 40 for each of the multiple moments in time by evaluating, for each of the multiple moments in time, one or more functions that map a set of one or more input parameters or variables including the corresponding lateral position of the second model vehicle O relative to the center of the lane 21C to a single output value of the corresponding lateral range.

[0122] Alternatively, in some example embodiments, the lateral range determination module 13 may determine the corresponding lateral range 40 for each of the plurality of moments by setting the lateral range 40 to a predetermined distance extending in the first lateral direction and in the second lateral direction from the lateral position of the second model vehicle O. That is, the corresponding lateral range 40 for each of the plurality of moments may extend along the y-axis on either side of the second model vehicle.

[0123] In such an example embodiment, the lateral range determination module 13 may determine the corresponding lateral range 40 for each of the plurality of moments, for example, using the corresponding lateral position of the second model vehicle O relative to the center of the lane 21C in which the second model vehicle O is located among the plurality of lanes, wherein the predetermined distance set for each of the plurality of moments is selected from a plurality of predetermined distances based on the corresponding lateral position of the second model vehicle O relative to the center of the lane 21C in which the second model vehicle O is located among the plurality of lanes. For example, in the case where the distance between the lateral position of the second model vehicle O and the center of the lane 21C in which the second model vehicle O is traveling is relatively large (e.g., when the second model vehicle O is switching lanes), a larger predetermined distance may be selected. Alternatively, the predetermined distance may have a fixed or constant value for each of the plurality of moments, and the corresponding lateral position of the second model vehicle O relative to the center of the lane 21C in which the second model vehicle O is located among the plurality of lanes may be used only to determine the lateral position from which the corresponding lateral range 40 should extend.

[0124] Typically, a corresponding lateral range 40 is determined for each of a plurality of moments by using the corresponding lateral position of the second model vehicle O relative to the center of a lane 21C in which the second model vehicle O is located among a plurality of lanes, so that the corresponding lateral range 40 increases as the distance between the lateral position of the second model vehicle O and the center of the lane 21C in which the second model vehicle O is traveling increases. The corresponding lateral range 40 can, for example, be defined to reach the next lane of the second model vehicle O when the second model vehicle O is performing a lane switch, so that the first model vehicle is prohibited from performing a simultaneous lane switch to the same lane.

[0125] Additionally or alternatively, in example embodiments such as the present example embodiment, where the lateral range determination module 13 determines the corresponding lateral range 40 for each of the plurality of moments in time by evaluating one or more functions, one or more input parameters or variables of the one or more functions may include any suitable parameters included in the second vehicle state.

[0126] As an example, in an example embodiment such as the present example embodiment, in which the second vehicle state for each of the plurality of moments is also defined by the corresponding lateral speed of the second model vehicle O, the lateral range determination module 13 may also be configured to use the corresponding lateral speed of the second model vehicle to determine the corresponding lateral range 40 for each of the plurality of moments. As an example, the lateral range 40 determined for a particular moment may be determined such that the lateral range 40 increases as the lateral speed of the second model vehicle O increases.

[0127] In the case where the second model vehicle O is moving at a relatively high lateral speed and thus may be approaching the first model vehicle relatively quickly, this may help ensure that the determined lateral range 40 used to determine whether a planned trajectory for the first model vehicle is valid is wider, and therefore more planned trajectories may be potentially invalid relative to the case where the second model vehicle O is moving at a relatively low lateral speed. In this way, it may be better ensured that any planned trajectory determined to be a valid planned trajectory allows the first model vehicle sufficient time and lateral distance to react in the case where the second model vehicle O is moving at a relatively high lateral speed.

[0128] As a more detailed example, in this example embodiment, the lateral range determination module 13 may first define the following functions and sets:

[0129] (2)clip(x,a,b)=min(max(a,x),b)

[0130] (3)

[0131] (4)ys=[-0.5,-Δy bias ,Δy bias ,0.5]

[0132] (5)rs=[1.5-Δy bias ,1,1-Δy bias ,1.5-Δy bias ],

[0133] and defining the following function for determining the left part and the right part of the lateral range to determine the corresponding lateral range 40 for each of the plurality of time instants:

[0134] (6)

[0135] in,

[0136] (7)

[0137]

[0138] (8)Δy range (y,v)=Δy base (y)+Δy v,range (y,v).

[0139] In particular, using function (8) above, the left portion of the lateral extent can be determined as Δy range (y i ,v y,i ), and the right part of the lateral range can be determined by antisymmetry as Δy range (-y i ,-v y,i ).

[0140] In the above equation (7), v μ is a threshold speed (e.g., a peak lateral speed for biasing when driving in the same lane) below which the lateral range does not increase in order to avoid an increase in lateral range caused by small oscillations of the second model vehicle O, and v min,switch is the minimum lateral speed used during or indicating a lane switch. By defining the functions for determining the left and right parts of the lateral extent 40 separately, it can be ensured that the extent of the lateral extent 40 increases only on one side in the speed direction of the lane.

[0141] Figure 5 is a graph illustrating the variation of the left range with the position in the lane, which is calculated for the object component using the lateral range function described herein to ensure that the component is constant outside the bias region whenever the vehicle is in the bias region and has a low lateral velocity. The range of lateral velocities v≤v μ Depend on Figure 5 The solid line in the figure shows that the range of lateral velocity v ≥ v min,switch Depend on Figure 5 The dashed line in is shown. In between, the linear interpolation between the two gives the value. Figure 5 As shown, the left range decreases when moving from center to the left, and the left range increases when moving from center to the right. When the speed to the left is higher, the left range will instead have an increasing range when moving left away from the center of the lane.

[0142] In the above sets (4) and (5) and functions (6) to (8), it is assumed that the y-axis is scaled to the lane width so that each increment or decrement of 1 on the y-axis represents the width of one lane, such as in the present exemplary embodiment, so that the value 0.5 can represent half the lane width of the road 20 (i.e., the distance between the center of the lane and the lane boundary). Alternatively, in an exemplary embodiment where the y-axis is not so scaled, the value 0.5 in sets (4) and (5) and functions (6) to (8) can be replaced with any suitable value corresponding to half the lane width of the road 20, and other values ​​(e.g., 1.5, 1, etc.) can be scaled accordingly.

[0143] Fig. 6A is a schematic illustration showing an example of a lateral range 40 determined by the lateral range determination module 13 of the apparatus 1 of Fig. 1. The lateral range 40 may represent a corresponding lateral range defined for a specific moment in time among a plurality of moments in time.

[0144] As in this example implementation, Fig. 6A The lateral extent 40 shown may be determined using functions (6) to (8) described above. Alternatively, the lateral extent 40 may be determined by any suitable means as described above.

[0145] exist Fig. 6A In the example shown, the lateral range 40 is used to define a region 50 extending around the second model vehicle O, the region extending not only laterally along the y-axis but also longitudinally along the x-axis. The longitudinal length of the region 50 can be defined, for example, by information obtained by the second vehicle state obtaining module 12 .

[0146] The second model vehicle state obtaining module 12 may be configured to only consider additional model vehicles within a predetermined longitudinal distance of the first model vehicle. Thus, the second vehicle state may only be obtained for the second model vehicle within such a predetermined longitudinal distance of the first model vehicle.

[0147] As in the present example embodiment, the apparatus 10 may also be arranged to determine a longitudinal extent extending from the second model vehicle O in a first longitudinal direction and in a second longitudinal direction opposite to the first longitudinal direction (i.e., along the x-axis on either side of the second model vehicle O) as the longitudinal length of the region 50. In such an example embodiment, the lateral extent determination module 13 may also be arranged to determine the longitudinal extent, or the apparatus 10 may include an additional module arranged to determine the longitudinal extent.

[0148] As an example, the determined longitudinal range may have a fixed value. For example, the fixed value may correspond to a maximum longitudinal safety distance, which is defined as the distance required for vehicle 1 to perform a longitudinal emergency maneuver such as braking to stop in time to avoid a collision with another vehicle in a scenario where vehicle 1 is traveling at a maximum permitted speed of vehicle 1 (e.g., a speed limit on a highway) on a shared road and assuming that the other vehicle stops instantaneously (e.g., when a collision is involved).

[0149] Alternatively, as in the present example embodiment, the longitudinal range may be determined by evaluating one or more functions that map a set of one or more input parameters or variables, eg, associated with the second model vehicle O, to a single output value of the longitudinal range.

[0150] Additionally or alternatively, as in this example embodiment, the apparatus 10 may be arranged to determine a respective longitudinal extent for each of the plurality of moments in time. Alternatively, the apparatus 10 may be arranged to determine a single longitudinal extent that is common to all of the plurality of moments in time.

[0151] Additionally or alternatively, as in the present example embodiment, the apparatus 10 may determine, for each of a plurality of moments in time, a corresponding first longitudinal extent extending from the second model vehicle O and in front of the second model vehicle O, and determine, for each of a plurality of moments in time, a corresponding second longitudinal extent extending from the second model vehicle O and behind the second model vehicle O. In this way, the region 50 may be defined such that its longitudinal extent in front of the second model vehicle O is different from its longitudinal extent behind the second model vehicle O. In this way, it may be considered how the respective directions of motion of the first model vehicle and the second model vehicle may affect the safe longitudinal distance required for the vehicle 1 to perform a longitudinal emergency maneuver such as timely braking to avoid a collision with the second model vehicle.

[0152] As an example, in an example embodiment such as the present example embodiment, in which the first vehicle state is further defined by the longitudinal position of the first model vehicle and the longitudinal speed of the first model vehicle, and the second vehicle state is further defined by the longitudinal speed of the second model vehicle O and the longitudinal acceleration of the second model vehicle O, as in the present example embodiment, the apparatus 10 may use the corresponding longitudinal speed of the first model vehicle, the corresponding longitudinal speed of the second model vehicle O, the corresponding longitudinal acceleration of the second model vehicle O, the predetermined reaction time of the second model vehicle O, the first predetermined longitudinal deceleration of the second model vehicle O, and the first predetermined longitudinal deceleration of the first model vehicle to determine, for each of the plurality of moments, a corresponding first longitudinal range extending from the second model vehicle O and in front of the second model vehicle O. The apparatus 10 may also use the corresponding longitudinal speed of the first model vehicle, the corresponding longitudinal speed of the second model vehicle O, the predetermined reaction time of the first model vehicle, the second predetermined longitudinal deceleration of the first model vehicle, and the second predetermined longitudinal deceleration of the second model vehicle O to determine, for each of the plurality of moments, a corresponding second longitudinal range extending from the second model vehicle O and behind the second model vehicle O.

[0153] As an example, as in the present example embodiment, the first predetermined longitudinal deceleration and the second predetermined longitudinal deceleration of the second model vehicle O may correspond to the minimum required emergency braking capability and the maximum achievable emergency braking response of the second model vehicle, respectively, so that the first predetermined longitudinal deceleration of the second model vehicle O has a smaller magnitude than the second predetermined longitudinal deceleration of the second model vehicle O. In addition, as in the present example embodiment, the first predetermined longitudinal deceleration and the second predetermined longitudinal deceleration of the first model vehicle may correspond to the maximum achievable emergency braking response and the minimum required emergency braking capability of the first model vehicle, respectively, so that the first predetermined longitudinal deceleration of the first model vehicle has a larger magnitude than the second predetermined longitudinal deceleration of the first model vehicle.

[0154] As a more detailed example, in this example embodiment, the device 10 may use the following function to determine the corresponding longitudinal range for each of the multiple moments:

[0155] (9)

[0156] (10)

[0157] Where Δx front Corresponding to the safe distance of vehicle 1 in front of other vehicles, t o,r is the reaction time of the other road user in seconds (s), b max is the vehicle 1 in ms -2 The maximum braking deceleration is in units, and b o,min is another road user in ms-2 The minimum emergency braking response is in units of . In addition, Δx back Corresponding to the safe distance of vehicle 1 behind other vehicles, t r is the reaction (adjustment) time of vehicle 1 in seconds (s), b o,max is another road user in ms -2 The maximum braking deceleration is in units, and b min is the minimum emergency braking response of vehicle 1 (ms -2 ).

[0158] In particular, using the above functions (9) and (19), the first longitudinal extent can be determined as Δx front (l,v x , l i , v x，i , a x，i ), and the second longitudinal extent can be determined as Δx back (l,v x , l i , v x，i ).

[0159] Figure 6B 1 is a schematic illustration showing an example of the lateral range 40 determined by the lateral range determination module 13 of the apparatus 10 of FIG. 1 at a specific moment when the second model vehicle O moves into the lane boundary area 23C. Figure 6B In the example of FIG. 1 , the corresponding lateral range 40 is determined for each of the plurality of moments in time according to equations (6) to (9) above, and is depicted by region 50. However, according to the above discussion, in some example embodiments, region 50 may not actually be determined by apparatus 10, in which case region 50 may be depicted only for the convenience of visualization. Figure 6B Shown in.

[0160] like Figure 6B As shown, although the second model vehicle O is in Fig. 6A The same lane 21C is in progress, but as shown in area 50, the range of the determined lateral range increases significantly in the direction of the lane 21B to which the second model vehicle O moves. That is, the lateral range 40 reaches the next lane of the second model vehicle O while the second model vehicle O is performing a lane switch, so that the first model vehicle can be prohibited from performing a simultaneous lane switch into the same lane 21B.

[0161] Additionally or alternatively, in an example embodiment such as the present example embodiment, in which the lateral range determination module 13 determines the left and right portions of the lateral range 40, as in the present example embodiment, the lateral range determination module 13 can determine the corresponding lateral range 40 for each of a plurality of moments so that the lateral range of the second model vehicle O is taken into account.

[0162] As an example, Figure 6C The lateral range determination module 13 determines Fig. 6A Schematic illustration of a first example of the left and right parts of the lateral extent 40. Figure 6C In the example of FIG. 1 , the left and right portions of the lateral extent are depicted by regions 51 and 52. However, according to the above discussion, in some example embodiments, regions 51 and 52 may not actually be determined by device 10, in which case these regions may be depicted merely for ease of visualization. Figure 6C Shown in.

[0163] exist Figure 6C In the example of , the left portion 51 of the lateral range extends from the rightmost edge of the second model vehicle O and to the left side of the second model vehicle O. Similarly, the right portion 52 of the lateral range extends from the leftmost edge of the second model vehicle O and to the right side of the second model vehicle O. Note that the terms "left side" and "right side" may more generally refer to a first lateral direction along the y-axis and a second lateral direction along the y-axis opposite to the first lateral direction.

[0164] Thus, in example embodiments such as the present example embodiment, where the left portion of the lateral extent is determined as Δy using function 8 above range (y i ,v y,i ) and the right side of the lateral range is determined as Δy range (-y i ,-v y,i ), the complete range of the left part 51 of the lateral range and the right part 52 of the lateral range can be respectively defined by the following ranges:

[0165] (11)

[0166] (12)

[0167] As an alternative, in an example embodiment such as the present example embodiment, in which the lateral range determination module 13 determines the left and right portions of the lateral range 40, the lateral range determination module 13 may not consider the lateral range of the second model vehicle O when determining the corresponding lateral range 40 for each of multiple moments in time.

[0168] As an example, Fig.6D The lateral range determination module 13 determines Fig. 6A A schematic illustration of a second example of the left and right portions of the lateral extent 40 of FIG. Fig.6D In the example of FIG. 5 , the left and right portions of the lateral extent are depicted by regions 53 and 54. However, according to the above discussion, in some example embodiments, regions 53 and 54 may not actually be determined by device 10, in which case these regions may be depicted merely for ease of visualization. Fig.6D Shown in.

[0169] exist Fig.6D In the example of , the left portion 53 of the lateral range extends from the lateral center of the second model vehicle O and to the left side of the second model vehicle O. Similarly, the right portion 54 of the lateral range extends from the lateral center of the second model vehicle O and to the right side of the second model vehicle O. Note that the terms "left side" and "right side" may more generally refer to a first lateral direction along the y-axis and a second lateral direction along the y-axis opposite to the first lateral direction.

[0170] Additionally or alternatively, as in the present example embodiment, the lateral range determination module 13 may also be arranged to determine any other lateral range useful in determining the validity of the obtained planned trajectory.

[0171] As an example, as in the present example embodiment, the lateral range determination module 13 may also be arranged to determine, for each of a plurality of moments in time, a respective second lateral range extending from the second model vehicle O. As in the present example embodiment, the second lateral range and the second longitudinal range (which may be defined, for example, using equation (10) above) may define a braking area of ​​the model of the road extending from the lateral center of the first model vehicle in the first lateral direction and the second lateral direction.

[0172] The respective second lateral extent may be determined by any of the means discussed above with respect to the respective lateral extent. As in this example embodiment, the respective second lateral extent may be determined using function (8) as described above with respect to the respective lateral extent determined by lateral extent determination module 13.

[0173] As another example, in an example embodiment such as the present example embodiment, where the first vehicle state is further defined by the longitudinal position of the first model vehicle and the second vehicle state is further defined by the length of the second model vehicle and the width of the second model vehicle, as in the present example embodiment, the lateral range determination module 13 may determine the corresponding occupied area occupied by the second model vehicle for each of the plurality of moments in time based on the length of the second model vehicle and the width of the second model vehicle. For example, as in the present example embodiment, the occupied area may correspond to Figure 2B and FIG. 6A to FIG. 6D The black rectangle represented by the reference numeral O in FIG.

[0174] Figure 7 is a schematic illustration showing an example of a braking zone 60 determined by the apparatus 10 of Figure 1 and represents an area where the vehicle 1 will be required to perform an emergency braking response. That is, if the vehicle 1 enters the area, then in the event that other road users in front of the vehicle 1 perform an emergency braking response, the vehicle 1 will need to apply full emergency braking to avoid a collision.

[0175] Reference again Figure 4 In the processing step S44 of the figure, the validity determination module 4 is configured to determine that the planned trajectory is invalid when: for the first vehicle state and the second vehicle state at one or more of the multiple moments, the following conditions in the first condition set are satisfied:

[0176] (i) the lateral position of the first model vehicle is within the lateral range 40 of the second model vehicle O at this moment and within the lane boundary area 23A, 23B, 23C, 23D, said lane boundary area 23A, 23B, 23C, 23D extending along and including the lane boundaries 22A, 22B, 22C, 22D among the plurality of lane boundaries; and

[0177] (ii) The direction of the lateral velocity of the first model vehicle is toward the second model vehicle O, and the magnitude of the lateral acceleration of the first model vehicle away from the second model vehicle O is smaller than the predetermined lateral deceleration of the first model vehicle.

[0178] As an example, the predetermined lateral deceleration may represent the minimum required lateral deceleration of the vehicle 1 (i.e., the acceleration of the first model vehicle away from the second model vehicle O). In this case, in the case where the lateral acceleration of the first model vehicle is toward the second model vehicle O and in the case where the lateral acceleration of the first model vehicle is away from the second model vehicle O (i.e., a deceleration), the magnitude of the lateral acceleration of the first model vehicle away from the second model vehicle O may be less than the predetermined lateral deceleration of the first model vehicle, but the deceleration of the first model vehicle is less than the predetermined lateral acceleration.

[0179] As a more specific example, as in the present embodiment, the area satisfying the above condition (i) at a given time t may be defined for the second model vehicle O as follows (wherein the subscript i indicates that the second model vehicle O may be the i-th model vehicle among one or more model vehicles defined in the model of the road 20 in addition to the first model vehicle):

[0180] (13)

[0181]

[0182] (14)

[0183]

[0184] Among them, R cutl refers to the right side portion of the lateral range (i.e., the portion intended to prevent the first model vehicle from cutting into the same lane as the second model vehicle O from the right side), R cutr refers to the left part of the lateral range (i.e., the part intended to prevent the first model vehicle from cutting into the same lane as the second model vehicle O from the left), R l are the lane boundary regions 23A, 23B, 23C, 23D defined in the above equation (1), H t is the first vehicle state at time t, E t is the environmental state at time t. According to equations (13) and (14), R cutl and R cutr YesR l , rather than the entire left or right portion of the horizontal extent.

[0185] Accordingly, as in the present embodiment, the set of acceptable states H of the first vehicle that satisfies the above condition (ii) at a given time t is t Can be targeted at R cutl and R cutr They are defined as follows:

[0186] (15)P cutl ={H t in H stv y <0 or a y <-b y,min}

[0187] (16)P cutr ={H t in H stv y >0 or a y >b y,min},

[0188] Among them, b v,min is the predetermined lateral deceleration.

[0189] As mentioned above about Figure 1A As described, Figure 4The process may allow the apparatus 10 to avoid invalidating a planned trajectory that represents an optimal response to unsafe driving by another road user, while ensuring that a planned trajectory that would result in the vehicle 1 being autonomously controlled to drive unsafely may be invalidated.

[0190] In addition, by obtaining the first vehicle state and the second vehicle state for each of multiple moments, and determining whether the planned trajectory is invalid when conditions (i) and (ii) are satisfied for even one of the multiple moments, the device 10 can allow the safety of the entire planned trajectory to be analyzed.

[0191] In some example embodiments such as the present example embodiment, in which the device 10 determines, for each of a plurality of moments, a corresponding first longitudinal range extending from the second model vehicle O and in front of the second model vehicle O and a corresponding second longitudinal range extending from the second model vehicle O and behind the second model vehicle O, as in the present example embodiment, the first condition set may also include a condition that the longitudinal position of the first model vehicle is within the first longitudinal range or the second longitudinal range at that moment.

[0192] Additionally or alternatively, in example embodiments such as the present example embodiment, in which the apparatus 10 determines a corresponding second lateral range extending from the second model vehicle O for each of a plurality of time instants, as in the present example embodiment, the validity determination module 14 may also be configured to determine that the planned trajectory is invalid if: for the first vehicle state and the second vehicle state at at least one of the plurality of time instants, the following conditions in the second condition set are satisfied:

[0193] (iii) the longitudinal position of the first model vehicle is within the second longitudinal range at this moment,

[0194] (iv) the lateral position of the first model vehicle is within the second lateral range at this moment, and

[0195] (v) The magnitude of the longitudinal acceleration of the first model vehicle moving away from the second model vehicle is less than a first predetermined longitudinal deceleration of the first model vehicle.

[0196] As an example, the first predetermined longitudinal deceleration may represent the minimum required longitudinal deceleration of the vehicle 1 (i.e., the acceleration of the first model vehicle away from the second model vehicle O). In this case, in the case where the lateral acceleration of the first model vehicle is toward the second model vehicle O and in the case where the lateral acceleration of the first model vehicle is away from the second model vehicle O (i.e., a deceleration), the magnitude of the lateral acceleration of the first model vehicle away from the second model vehicle O may be less than the predetermined lateral deceleration of the first model vehicle, but the deceleration of the first model vehicle is less than the predetermined lateral acceleration.

[0197] As a more specific example, as in the present embodiment, the area satisfying the above conditions (iii) and (iv) at a given time t may be defined for the second model vehicle O as follows (wherein the subscript i indicates that the second model vehicle O may be the i-th model vehicle among one or more model vehicles defined in the model of the road 20 in addition to the first model vehicle):

[0198] (17)

[0199]

[0200] Among them, R b Refers to the braking area, H t is the first vehicle state at time t, and E t is the state of the environment at time t.

[0201] Accordingly, as in the present embodiment, the set of acceptable states H of the first vehicle that satisfies the above condition (v) at a given time t is t It can be defined as follows:

[0202] (18)P b ={H t in H sta x ≤-b min}

[0203] Among them, b min is the first predetermined longitudinal deceleration.

[0204] Thus, the second set of conditions may allow the validity determination module 14 to determine that the planned trajectory is invalid if the planned trajectory would result in the vehicle 1 being autonomously controlled into a braking zone without appropriate braking.

[0205] Additionally or alternatively, in example embodiments such as the present example embodiment, in which the apparatus 10 determines a corresponding occupied area for each of a plurality of moments, as in the present example embodiment, the validity determination module 14 may also be configured to determine that the planned trajectory is invalid when: for the first vehicle state and the second vehicle state at at least one of the plurality of moments, the following conditions in the third condition set are satisfied:

[0206] (vi) The longitudinal position of the first model vehicle and the lateral position of the first model vehicle are within the occupied area occupied by the second model vehicle at this moment.

[0207] As a specific example, as in the present embodiment, the area satisfying the above condition (vi) at a given time t may be defined for the second model vehicle O as follows (wherein the subscript i indicates that the second model vehicle O may be the i-th model vehicle among one or more model vehicles defined in the model of the road 20 in addition to the first model vehicle):

[0208] (19)

[0209] Among them, H t is the first vehicle state at time t, and E t is the state of the environment at time t.

[0210] Thus, the third set of conditions may allow the validity determination module 14 to determine that the planned trajectory is invalid if the planned trajectory would cause the vehicle 1 to collide with the vehicle modeled by the second model vehicle. Since the vehicle 1 may never collide with another vehicle, the set of allowed responses to entering the occupied zone is an empty set. That is, there is no appropriate mitigation response, and such a trajectory must always be determined to be invalid.

[0211] Additionally or alternatively, as in the present example embodiment, the validity determination module 14 may also be configured to determine that the planned trajectory is invalid in the following case: at an initial time in the multiple time instants and at a last time in the multiple time instants, the lateral position of the first model vehicle is located in a common lane in the multiple lanes, and for the first vehicle state at at least one intermediate time in the multiple time instants, the following conditions in the fourth condition set are satisfied:

[0212] (vii) The lateral position of the first model vehicle is, at at least one intermediate moment, within a lane boundary region 23A, 23B, 23C, 23D that extends along and includes the lane boundaries of the common lane.

[0213] As an example, the at least one intermediate time may be between the initial time and the final time.

[0214] In particular, as discussed in detail above, the vehicle 1 should only be autonomously controlled to enter the lane boundary regions 23A, 23B, 23C, 23D when it is expected to switch lanes. As such, unless the first model vehicle is switching lanes, a planned trajectory that would cause the vehicle 1 to be autonomously controlled to enter the lane boundary regions 23A, 23B, 23C, 23D without causing a lane switch may be considered unsafe because the first model vehicle should stay near the center of the lane being traveled, particularly at the maximum lateral deviation Δy from the center of the lane. bias Inside.

[0215] Thus, the fourth condition set may allow the validity determination module 14 to determine that the planned trajectory is invalid if the planned trajectory would cause the vehicle 1 to enter the lane boundary region 23A, 23B, 23C, 23D if the planned trajectory did not result in a lane switch.

[0216] In example embodiments such as the present example embodiment, where the validity determination module 14 determines whether the planned trajectory is valid based on multiple sets of conditions, the validity determination module 14 may determine that the planned trajectory is invalid based on one or more of the sets of conditions used. In this case, as in the present example embodiment, the validity determination module 14 may be configured to determine that the planned trajectory is invalid, where the planned trajectory is determined to be invalid based on any one of the sets of conditions.

[0217] As in the present example embodiment, the validity determination module 14 may be configured to determine that the planned trajectory is a valid planned trajectory if the planned trajectory is not determined to be invalid.

[0218] As a specific example, in this example implementation, the condition for the validity determination module 14 to determine that the planned trajectory is a valid planned trajectory can be defined in the following form:

[0219] (20)

[0220] Where S is equal to o, b, cutl, cutr, T is multiple time t, H t is the first vehicle state at time t, E t is the state of the environment at time t, and (x, y) refers to the position in the trajectory at time t.

[0221] In particular, it is apparent that in example embodiments such as the present example embodiment, in which the validity determination module 14 determines whether a planned trajectory is valid based on multiple sets of conditions, the various areas defined by the condition sets may overlap, so that the planned trajectory may cause the position of the first model vehicle to be located in two or more areas at one or more of the multiple moments in time.

[0222] Figure 8 is a diagram showing an exemplary embodiment of the present invention. Figure 4 Schematic illustration of an example of all regions defined by the process. Figure 8 In the example, the lateral range is depicted by area 50, the braking area is indicated by reference symbol 60, the occupied area is indicated by reference symbol O, the lane boundary areas are indicated by reference symbols 23A, 23B, 23C, 23D, and the overlapping areas between area 50 and the lane boundary areas 23A, 23B, 23C, 23D are indicated by reference symbols 70 and 71.

[0223] In some example embodiments, such as in this example embodiment, Figure 4 The process may optionally include an additional process step, wherein the validity determination module 14 outputs the valid planned trajectory to the autonomous driving system 15 so as to enable the autonomous driving system 15 to autonomously control the vehicle to drive along the road according to the valid planned trajectory. Alternatively, the apparatus 10 may optionally include an additional module configured to output the valid planned trajectory to the autonomous driving system 15.

[0224] As another alternative, where the apparatus 10 is used to determine whether a trajectory is invalid, for example as part of a simulation of an autonomously controlled vehicle or as part of testing and analysis of an autonomous driving algorithm (such as an autonomous driving algorithm based on a statistical model), the validity determination module 14 may be configured to output the valid planned trajectory to any suitable entity used in the testing and analysis of the autonomous driving algorithm, such as a connected computing device or smart phone.

[0225] Additionally or alternatively, in some example embodiments, the apparatus 10 may be arranged to receive a plurality of planned trajectories, the validity of which may be determined using Figure 4 In such an example embodiment, when no planned trajectory is determined as a valid trajectory, the device 10 may also be arranged to output a notification to the entity responsible for generating the planned trajectory to prompt the planned trajectory to be generated again. Additionally or alternatively, the device 10 may also be arranged to output a notification or warning that no valid planned trajectory is determined to any suitable entity.

[0226] In the aforementioned example embodiment, the first model vehicle and the second model vehicle O are each represented as an object having dynamic characteristics defined in a lane coordinate system. As an alternative, the first model vehicle and the second model vehicle may be represented as one or more cells of a grid having dynamic characteristics defined in a lane coordinate system. For example, one or more cells of the grid may correspond to, for example, a bounding box or other area occupied by the second model vehicle. In this way, the outermost cells or points in each direction may be considered to define the scope of the bounding box or other area occupied by the second model vehicle in that direction. The first model vehicle and any additional model vehicles may be defined in a grid in a similar manner.

[0227] As an example, Fig.9A and Fig. 9B is a diagram showing that for each of the first unit O_1 and the second unit O_n in the plurality of units representing the second model vehicle O, Figure 4 Schematic illustration of an example of all regions defined by the process. Fig.9AIn the example, the first unit O_1 is the unit farthest in the positive longitudinal direction (+x) and the positive direction (+y) among multiple units representing the second model vehicle O, and the second unit O_n is the unit farthest in the negative longitudinal direction (-x) and the negative lateral direction (-y) among multiple units representing the second model vehicle O.

[0228] It should be noted that the lateral extent determined and the above Figure 4 The other ranges described in the process are, for example, defined relative to the center of the bounding box of the second model vehicle. Alternatively, in an example embodiment in which the first model vehicle and the second model vehicle are represented as one or more cells of a grid having dynamic characteristics defined in a lane coordinate system, the lateral range determination unit 13 of the device 10 can be configured to determine the corresponding lateral ranges of some or all of the multiple cells of the bounding box representing the second model vehicle O for each of the multiple moments.

[0229] At any given moment, when the second model vehicle O is represented as an object with dynamic characteristics defined in the lane coordinate system, the union of the corresponding lateral ranges of the individual cells in the cells that calculate the lateral range can be equivalent to the single lateral range calculated at that moment. More specifically, when the second model vehicle O is represented as an object with dynamic characteristics defined in the lane coordinate system, the union of the corresponding lateral ranges of the individual cells in the cells that calculate the lateral range can be equivalent to the single lateral range calculated at that moment, with an error tolerance depending on the grid resolution. Therefore, the lateral range determination unit 13 of the device 10 can be configured to determine the single lateral range at a given moment as the union of the corresponding lateral ranges of the individual cells in the cells that calculate the lateral range for that moment.

[0230] As an example, in an example embodiment such as the present example embodiment, where the lateral range determination unit 13 uses the above function (8) to determine the lateral range, note that for the lateral velocity v of the second model vehicle O y,i For all values ​​of , function (8) changes with the lateral position y of the second model vehicle O i Monotonically increasing. In this way, the lateral range determined for the middle cell among the multiple cells representing the second model vehicle O will not extend further than the lateral range determined for the outermost cell. Therefore, in the case where the second model vehicle O is represented as an object with dynamic characteristics defined in the lane coordinate system, the union of the lateral ranges of the outermost cell points of a given object will be equivalent to a single lateral range determined at that moment.

[0231] For example, Fig.9A and Fig. 9BAs shown, in the case where the second model vehicle O is represented as an object with dynamic characteristics defined in the lane coordinate system, the union of the lateral ranges (and other ranges and areas) determined for each unit O_1 and O_n in the unit is equivalent to the determined Figure 8 A single lateral extent (as well as other extents and regions) is shown.

[0232] The model of the road 20 may be obtained by any suitable means known to those skilled in the art, wherein the environment of the first model vehicle (particularly the first model vehicle and any additional model vehicles) is represented as one or more cells of a grid having dynamic characteristics defined in a suitable coordinate system. As an example, each of the following papers defines a technique by which the environment of a vehicle may be represented as a grid defined in a suitable coordinate system, the grid having occupied cells, the cells having dynamic characteristics:

[0233] ·Dominik Nuss et al.: "A Random Finite Set Approach for DynamicOccupancy Grid Maps with Real-Time Application", (arXiv:1605.02406v2);

[0234] ·S.Steyer, et al.: "Grid-Based Environment Estimation Using EvidentialMapping and Particle Tracking," in IEEE Transactions on Intelligent Vehicles, vol.3, no.3, pp.384-396, Sept.2018, doi:10.1109 / TIV.2018.2843130;

[0235] ·Stefan Hoermann et al,: "Dynamic Occupancy Grid Prediction for UrbanAutonomous Driving: A Deep Learning Approach with Fully Automatic Labeling", (arXiv:1705.08781v2); and

[0236] ·Christopher Diehl et al.: "Radar-based Dynamic Occupancy GridMapping and Object Detection", (arXiv:2008.03696v1).

[0237] The example aspects described herein avoid limitations associated with the field of autonomous driving, particularly based on computer technology. With the help of the example aspects described herein, it is possible to avoid determining that a planned trajectory representing the best response to unsafe driving of another road user is invalid, while ensuring that a planned trajectory that would cause the vehicle to be autonomously controlled to drive unsafely can be invalidated. In addition, by obtaining a first vehicle state and a second vehicle state for each of a plurality of moments, and determining whether the planned trajectory is invalid when at least a first set of conditions is satisfied for even one of the plurality of moments, the technology herein can allow the safety of the entire planned trajectory to be analyzed. In some example embodiments such as the present example embodiment, wherein the device 10 determines a corresponding first longitudinal range extending from the second model vehicle O and in front of the second model vehicle O and a corresponding second longitudinal range extending from the second model vehicle O and behind the second model vehicle O for each of the plurality of moments, as in the present example embodiment, the first condition set may also include a condition that the longitudinal position of the first model vehicle is in the first longitudinal range or the second longitudinal range at that moment. In addition, with the aforementioned capabilities of the computer technology-based example aspects described herein, the example aspects described herein improve computers and computer processing / functionality, and also improve at least the field of autonomous driving, and in particular, improve determining whether a planned trajectory of a first model vehicle used in autonomous control of a vehicle modeled by the first model vehicle is invalid.

[0238] In the foregoing description, various aspects have been described with reference to several embodiments. Therefore, the description should be considered illustrative rather than restrictive. Similarly, the figures illustrated in the accompanying drawings highlighting the functions and advantages of the embodiments are presented for illustrative purposes only. The architecture of the embodiments is sufficiently flexible and configurable so that it can be utilized in a manner different from that shown in the accompanying drawings.

[0239] In an exemplary embodiment, the software implementation presented herein may be provided as a computer program or software, such as one or more programs with instructions or instruction sequences, which are included or stored in products such as machine-accessible media or machine-readable media, instruction storage units or computer-readable storage devices, each of which may be non-transitory. Programs or instructions on non-transitory machine-accessible media, machine-readable media, instruction storage units or computer-readable storage devices can be used to program computer systems or other electronic devices. Machine-readable media or computer-readable media, instruction storage units and storage devices may include, but are not limited to, floppy disks, optical disks and magneto-optical disks or other types of media / machine-readable media / instruction storage units / storage devices suitable for storing or transmitting electronic instructions. The techniques described herein are not limited to any particular software configuration. They can find applicability in any computing or processing environment. The terms "computer-readable medium", "machine-accessible medium", "machine-readable medium", "instruction storage unit" and "computer-readable storage device" used herein should include any medium capable of storing, encoding or transmitting instructions or instruction sequences for execution by a machine, computer or computer processor and causing the machine / computer / computer processor to perform any of the methods described herein. Furthermore, it is common in the art to refer to software in one form or another (e.g., program, procedure, process, application, module, unit, logic, etc.) as taking an action or causing a result. Such expressions are merely a shorthand way of stating that execution of the software by a processing system causes the processor to perform an action to produce a result.

[0240] Some embodiments may also be implemented by the preparation of application specific integrated circuits, field programmable gate arrays, or by interconnecting an appropriate network of conventional component circuits.

[0241] Some embodiments include computer program products. A computer program product may be a storage medium or multiple storage media, an instruction storage unit or multiple instruction storage units, or a storage device or multiple storage devices having instructions stored thereon or therein, which instructions may be used to control or cause a computer or computer processor to perform any program of the example embodiments described herein. Storage media / instruction storage units / storage devices may include, for example but not limited to, optical disks, ROMs, RAMs, EPROMs, EEPROMs, DRAMs, VRAMs, flash memories, flash cards, magnetic cards, optical cards, nanosystems, molecular memory integrated circuits, RAIDs, remote data storage units / archiving / warehousing, and / or any other type of device suitable for storing instructions and / or data.

[0242] Some implementations stored on a computer-readable medium or multiple media, an instruction store or multiple instruction stores, or a storage device or multiple storage devices include hardware for controlling the system and software for enabling the system or microprocessor to interact with a human user or other mechanism using the results of the embodiments described herein. Such software may include, but is not limited to, device drivers, operating systems, and user applications. Finally, such computer-readable media or storage devices also include software for performing the example aspects described above.

[0243] Included in the programming and / or software of the system are software modules for implementing the processes described herein. In some example embodiments herein, the modules include software, while in other example embodiments herein, the modules include hardware or a combination of hardware and software.

[0244] Although various embodiments of the present disclosure have been described above, it should be understood that they are presented as examples rather than limitations. It will be apparent to those skilled in the relevant art that various changes in form and detail may be made therein. Therefore, the present disclosure should not be limited by any of the above-described example embodiments, but should only be limited according to the attached claims and their equivalents.

[0245] In addition, the purpose of the abstract is to enable patent offices and the public, especially scientists, engineers and practitioners in the field who are not familiar with patent or legal terms or wording, to quickly determine the nature and substance of the technical disclosure of the present application from a cursory inspection. The abstract is not intended to limit the scope of the embodiments presented herein in any way. It should also be understood that any process recited in the claims need not be performed in the order presented.

[0246] Although this specification contains many specific implementation details, these details should not be interpreted as limitations on the content claimed, but as descriptions of specific features of specific implementations described herein. Certain features described in the context of separate implementations in this specification may also be implemented in combination in a single implementation. On the contrary, the various features described in the context of a single implementation may also be implemented individually or in any suitable sub-combination in multiple implementations. In addition, although features may be described above as working in certain combinations and even initially claimed as such, one or more features from the claimed combination may be deleted from the combination in some cases, and the claimed combination may be directed to a sub-combination or a variation of the sub-combination.

[0247] In some cases, multitasking and parallel processing may be advantageous. In addition, the separation of various components in the above-mentioned embodiments should not be understood as requiring such separation in all embodiments, and it should be understood that the described program components and systems can generally be integrated together in a single software product or packaged into multiple software products.

[0248] Now that some exemplary embodiments have been described, it is apparent that the foregoing is illustrative and not restrictive and has been presented by way of example. In particular, although many of the examples presented herein relate to specific combinations of devices or software elements, these elements may be combined in other ways to achieve the same purpose. Actions, elements, and features discussed only in conjunction with one embodiment are not intended to be excluded from similar effects in other embodiments or multiple embodiments.

[0249] The devices described herein may be implemented in other specific forms without departing from the features of the devices described herein. The above embodiments are illustrative rather than limiting of the described systems and methods. Therefore, the scope of the devices described herein is indicated by the appended claims rather than the preceding description, and changes falling within the meaning and range of equivalents of the claims are included therein.

Claims

1. A computer-implemented method for determining whether a planned trajectory of a first model vehicle on a model of a road (20) along which the first model vehicle and a second model vehicle (O) are traveling is invalid, the planned trajectory being used for autonomous control of a vehicle (1) modeled by the first model vehicle, the model of the road (20) comprising a plurality of lanes defined by a plurality of lane boundaries, the method comprising the following steps: Obtaining (S41) the planned trajectory (24), the planned trajectory (24) comprising a corresponding first vehicle state of the first model vehicle defined by the model relative to the road (20) for each of a plurality of moments in time, each first vehicle state being defined by a corresponding lateral position of the first model vehicle, a corresponding lateral velocity of the first model vehicle and a corresponding lateral acceleration of the first model vehicle; Obtaining (S42) a second vehicle state of the second model vehicle (O) defined relative to the model of the road (20) for each of the plurality of moments, each second vehicle state being defined by a corresponding lateral position of the second model vehicle (O) in the model of the road (20); determining (S43) a respective lateral range extending from the second model vehicle (O) for each of the plurality of moments in time using the respective lateral positions of the second model vehicle (O) relative to the center of the lane in which the second model vehicle (O) is located among the plurality of lanes; and It is determined (S44) that the planned trajectory (24) is invalid in the following case: for the first vehicle state and the second vehicle state at at least one of the plurality of time instants, the following conditions in a first condition set are satisfied: The lateral position of the first model vehicle at the time is within the lateral range of the second model vehicle (O) and within a lane boundary area, the lane boundary area extending along and including a lane boundary among the plurality of lane boundaries; and The direction of the lateral velocity of the first model vehicle is toward the second model vehicle (O), and the magnitude of the lateral acceleration of the first model vehicle away from the second model vehicle (O) is less than a predetermined lateral deceleration of the first model vehicle.

2. The method according to claim 1, wherein: The second vehicle state for each of the plurality of moments in time is also defined by a corresponding lateral velocity of the second model vehicle (O), and The respective lateral extents are determined using the respective lateral velocities of the second model vehicle (O) for each of the plurality of moments in time.

3. The method according to claim 1 or 2, wherein: The first vehicle state is also defined by the longitudinal position of the first model vehicle and the longitudinal speed of the first model vehicle, The second vehicle state is also defined by the longitudinal velocity of the second model vehicle (O) and the longitudinal acceleration of the second model vehicle (O), and The method further comprises the following steps: determining, for each of the plurality of moments in time, a respective first longitudinal extent extending from and in front of the second model vehicle (O) using the respective longitudinal speed of the first model vehicle, the respective longitudinal speed of the second model vehicle (O), the respective longitudinal acceleration of the second model vehicle (O), a predetermined reaction time of the second model vehicle, a first predetermined longitudinal deceleration of the second model vehicle (O), and a first predetermined longitudinal deceleration of the first model vehicle; and A corresponding second longitudinal range extending from and behind the second model vehicle (O) is determined for each of the plurality of moments using the corresponding longitudinal speed of the first model vehicle, the corresponding longitudinal speed of the second model vehicle (O), a predetermined reaction time of the first model vehicle, a second predetermined longitudinal deceleration of the first model vehicle, and a second predetermined longitudinal deceleration of the second model vehicle (O).

4. The method according to claim 3, wherein: The first condition set further includes the following condition: the longitudinal position of the first model vehicle is within the first longitudinal range or the second longitudinal range at the moment.

5. The method according to claim 3 or 4, further comprising the following steps: determining, for each of the plurality of moments in time, a respective second lateral extent extending from the second model vehicle (O), wherein the second lateral extent and the second longitudinal extent define a braking region of the model of the road (20) extending from a lateral center of the first model vehicle in a first lateral direction and a second lateral direction; and The planned trajectory (24) is determined to be invalid when: for the first vehicle state and the second vehicle state at at least one of the plurality of time instants, the following conditions in a second set of conditions are satisfied: The longitudinal position of the first model vehicle is within the second longitudinal range at the moment, the lateral position of the first model vehicle is within the second lateral range at the moment, and the amplitude of the longitudinal acceleration of the first model vehicle away from the second model vehicle (O) is less than the first predetermined longitudinal deceleration of the first model vehicle.

6. A method according to any one of the preceding claims, wherein: The first vehicle state is also defined by the longitudinal position of the first model vehicle, The second vehicle state is also defined by the length of the second model vehicle (O) and the width of the second model vehicle (O), and The method further comprises the following steps: determining, for each of the plurality of moments in time, a corresponding occupied area occupied by the second model vehicle (O) based on the length of the second model vehicle (O) and the width of the second model vehicle (O); The planned trajectory (24) is determined to be invalid when: for the first vehicle state and the second vehicle state at at least one of the plurality of time instants, the following conditions in a third condition set are satisfied: The longitudinal position of the first model vehicle and the lateral position of the first model vehicle are within the occupied area occupied by the second model vehicle (O) at the moment.

7. A method according to any one of the preceding claims, wherein: The second model vehicle (O) is defined by: coordinates identifying the position of the second model vehicle (O) in a lane coordinate system of the model of the road (20), the length of the second model vehicle (O) along the longitudinal axis of the lane coordinate system, and the width of the second model vehicle (O) along the transverse axis of the lane coordinate system, or One or more cells of a grid defined in the lane coordinate system.

8. The method according to any one of the preceding claims, further comprising the steps of: The planned trajectory (24) is determined to be invalid when: at an initial moment in the plurality of moments and at a final moment in the plurality of moments, the lateral position of the first model vehicle is located in a common lane in the plurality of lanes, and for the first vehicle state at at least one intermediate moment in the plurality of moments, the following conditions in a fourth set of conditions are satisfied: The lateral position of the first model vehicle at the at least one intermediate time is within a lane boundary region that extends along and includes a lane boundary of the common lane.

9. A method according to any one of the preceding claims, wherein: A time interval between each pair of consecutive moments in the plurality of moments is less than or equal to a predetermined reaction time of the first model vehicle.

10. The method according to any one of the preceding claims, further comprising the steps of: If the planned trajectory (24) is not determined to be invalid, determining that the planned trajectory (24) is a valid planned trajectory; as well as The effective planned trajectory is output to the autonomous driving system (15) so that the autonomous driving system (15) can autonomously control the vehicle (1) to drive along the road (20) according to the effective planned trajectory.

11. A computer program (345) comprising computer program instructions which, when executed by a computer processor (320), cause the computer processor (320) to perform the method according to at least one of the preceding claims.

12. An apparatus (10) arranged to determine whether a planned trajectory (24) of a first model vehicle on a model of a road (20) along which the first model vehicle and a second model vehicle (O) are travelling is invalid, the planned trajectory (24) being used for autonomous control of a vehicle (1) modelled by the first model vehicle, the model of the road (20) comprising a plurality of lanes defined by a plurality of lane boundaries, the apparatus (10) comprising: a planned trajectory acquisition module (11), the planned trajectory acquisition module being configured to obtain the planned trajectory (24), the planned trajectory (24) comprising a corresponding first vehicle state of the first model vehicle defined by the model relative to the road (20) for each of a plurality of moments, each first vehicle state being defined by a corresponding lateral position of the first model vehicle, a corresponding lateral velocity of the first model vehicle and a corresponding lateral acceleration of the first model vehicle; a second vehicle state obtaining module (12), the second vehicle state obtaining module being configured to obtain a second vehicle state of the second model vehicle (O) defined relative to the model of the road (20) for each of the plurality of moments, each second vehicle state being defined by a corresponding lateral position of the second model vehicle (O) in the model of the road (20); a lateral range determination module (13) configured to determine, for each of the plurality of moments in time, a respective lateral range extending from the second model vehicle (O) using the respective lateral position of the second model vehicle (O) relative to a center of a lane of the plurality of lanes in which the second model vehicle (O) is located; as well as A validity determination module (14) is configured to determine that the planned trajectory (24) is invalid when: for the first vehicle state and the second vehicle state at at least one of the plurality of time instants, the following conditions in a first condition set are satisfied: The lateral position of the first model vehicle at the time is within the lateral range of the second model vehicle (O) and within a lane boundary area, the lane boundary area extending along and including a lane boundary among the plurality of lane boundaries; and The direction of the lateral velocity of the first model vehicle is toward the second model vehicle (O), and the magnitude of the lateral acceleration of the first model vehicle away from the second model vehicle (O) is less than a predetermined lateral deceleration of the first model vehicle.

13. A vehicle (1), comprising: Autonomous driving systems (15); as well as The apparatus (10) according to claim 12, the apparatus being arranged to determine whether a planned trajectory (24) is invalid, the apparatus being further arranged to determine that the planned trajectory (24) is a valid planned trajectory if the planned trajectory (24) is not determined to be invalid, and outputting the valid planned trajectory to the autonomous driving system (15), The autonomous driving system (15) is arranged to autonomously control the vehicle (1) to drive along a road (20) according to the effective planned trajectory.

Images