Track planning method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention provides a trajectory planning method and device, electronic equipment and a storage medium, and relates to the technical field of automatic driving, and the method comprises the steps: obtaining a plurality of pieces of prediction trajectory information corresponding to each target obstacle, the target obstacle comprises an obstacle in a preset range of the vehicle; based on the multiple pieces of prediction trajectory information corresponding to the target obstacle, multiple pieces of boundary information corresponding to the target obstacle at each future moment are determined, and the multiple pieces of boundary information corresponding to the future moments are determined based on the multiple prediction trajectory points corresponding to the future moments; determining a plurality of pieces of planning track information of the vehicle based on the plurality of pieces of boundary information corresponding to the target obstacle at each future moment; one piece of planning track information is selected from the multiple pieces of planning track information to serve as the target planning track information of the vehicle, and therefore the accuracy of track planning of the vehicle can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automatic driving, and more particularly, to a trajectory planning method and device, an electronic device, and a storage medium. BACKGROUND

[0002] At present, cruise systems are gradually popularized in the automobile market. Common cruise systems include adaptive cruise control (ACC) systems, intelligent cruise assist (ICA) systems, and the like. Cruise systems can control vehicles to cruise at a set speed or a following distance when a driver does not control the vehicle, which can bring great convenience to the driver. In addition, in order to ensure safety, it is usually necessary to predict the behavior of an interfering vehicle (also referred to as an obstacle), determine whether the prediction result will affect the safety distance set by the ego vehicle, and then make a corresponding decision and speed planning, so as to ensure the safety of driving.

[0003] In related technologies, the trajectory information of an obstacle is predicted, and then the trajectory information of the ego vehicle is planned in combination with the predicted trajectory information of the obstacle.

[0004] However, long-term research has found that the trajectory information of the ego vehicle planned in related technologies is not accurate enough, which can cause a sudden change in the planned response behavior. SUMMARY

[0005] Embodiments of the present application provide a trajectory planning method and device, an electronic device, and a storage medium, which can improve the accuracy of trajectory planning and reduce sudden changes in the planned response behavior.

[0006] In a first aspect, embodiments of the present application provide a trajectory planning method, which includes: obtaining a plurality of pieces of predicted trajectory information corresponding to each target obstacle, any piece of predicted trajectory information including a plurality of predicted trajectory points corresponding to respective future time points, and the target obstacle including an obstacle within a preset range of an ego vehicle; determining a plurality of pieces of boundary information corresponding to the target obstacle at respective future time points based on the plurality of pieces of predicted trajectory information corresponding to the target obstacle, the plurality of pieces of boundary information corresponding to the future time points being determined based on the plurality of predicted trajectory points corresponding to the future time points; determining a plurality of pieces of planned trajectory information of the ego vehicle based on the plurality of pieces of boundary information corresponding to the target obstacle at the respective future time points, any piece of planned trajectory information including a plurality of planned trajectory points corresponding to respective future time points; and selecting one of the plurality of pieces of planned trajectory information as target planned trajectory information of the ego vehicle.

[0007] In this embodiment, multiple predicted trajectory information corresponding to each target obstacle is acquired. Each predicted trajectory information includes multiple predicted trajectory points corresponding to each future time. The target obstacle includes obstacles within a preset range of the vehicle. Based on the multiple predicted trajectory information corresponding to the target obstacle, multiple boundary information corresponding to the target obstacle at each future time is determined. The multiple boundary information corresponding to each future time is determined based on the multiple predicted trajectory points corresponding to each future time. Based on the multiple boundary information corresponding to the target obstacle at each future time, multiple planned trajectory information of the vehicle is determined. Each planned trajectory information includes multiple planned trajectory points corresponding to each future time. One of the multiple planned trajectory information is selected as the target planned trajectory information of the vehicle. In this way, the trajectory planning of the vehicle can be performed by combining multiple predicted trajectory information of the target obstacle. Since the trajectory planning is performed by combining multiple predicted trajectory information, more information can be referenced for predicting the trajectory, thereby improving the accuracy of the vehicle's trajectory planning.

[0008] In one possible implementation, the boundary information includes a first boundary and a second boundary. The distance between the first boundary and the current position of the vehicle is less than the distance between the second boundary and the current position of the vehicle. Based on multiple predicted trajectory information corresponding to the target obstacle, multiple boundary information corresponding to the target obstacle at each future moment is determined, including: generating a spatiotemporal map based on the multiple predicted trajectory information corresponding to the target obstacle. The spatiotemporal map is used to represent multiple occupied areas corresponding to the target obstacle at each of the multiple future moments. Each occupied area is determined based on a predicted trajectory point in one of the predicted trajectory information. The occupied area is located in the direction of travel of the vehicle. For any future moment, a second boundary is determined based on the overlapping part between the multiple occupied areas corresponding to the future moment, and a first boundary is determined based on the non-overlapping part corresponding to the future moment. The non-overlapping part includes the part other than the overlapping part in the total area formed by the multiple occupied areas.

[0009] In one possible implementation, multiple planned trajectory information of the vehicle is determined based on multiple boundary information corresponding to the target obstacle at each future time. This includes: for the first future time among multiple future times, determining at least one planned trajectory point corresponding to the first future time based on the vehicle's current position and the multiple boundary information corresponding to the target obstacle at the first future time; for the Lth future time among multiple future times, determining at least one planned trajectory point corresponding to the Lth future time based on the planned trajectory point corresponding to the vehicle at the (L-1)th future time and the multiple boundary information corresponding to the target obstacle at the Lth future time, where L is an integer not less than 2; and determining multiple planned trajectory information of the vehicle based on the planned trajectory points corresponding to each of the multiple future times.

[0010] In one possible implementation, the predicted trajectory information also includes multiple first velocities corresponding to each future time. Based on the planned trajectory point of the vehicle at the (L-1)th future time and multiple boundary information of the target obstacle at the Lth future time, at least one planned trajectory point corresponding to the Lth future time is determined, including: based on the planned trajectory point of the vehicle at the (L-1)th future time, the second velocity of the vehicle at the (L-1)th future time, the multiple boundary information of the target obstacle at the Lth future time, and the multiple first velocities of the target obstacle at the L-1th future time, multiple second decelerations of the vehicle at the L-1th future time are determined, with the multiple boundary information, multiple first velocities, and multiple second decelerations corresponding one-to-one; the multiple second decelerations of the vehicle at the L-1th future time are fused to obtain a fused second deceleration; based on the fused second deceleration of the vehicle at the L-1th future time, a safe area of ​​the vehicle at the Lth future time is determined; and based on the points within the safe area of ​​the vehicle at the Lth future time, at least one planned trajectory point corresponding to the Lth future time is determined.

[0011] In one possible implementation, multiple second decelerations corresponding to the vehicle at the (L-1)th future time are fused to obtain the fused second deceleration. This includes: obtaining the weights corresponding to each second deceleration, where the weights are related to the prediction probability of the predicted trajectory information where the first velocity corresponding to the second deceleration is located; and performing a weighted calculation on the multiple second decelerations corresponding to the vehicle at the (L-1)th future time based on the weights corresponding to each second deceleration to obtain the fused second deceleration.

[0012] In one possible implementation, the predicted trajectory information further includes first decelerations corresponding to multiple future time points. Before fusing the multiple second decelerations corresponding to the vehicle at the (L-1)th future time point to obtain the fused second deceleration, the method further includes: obtaining the first deceleration of the target obstacle at the (L-1)th future time point; correcting the second deceleration of the vehicle at the (L-1)th future time point based on the first deceleration of the target obstacle at the (L-1)th future time point to obtain the corrected second deceleration of the vehicle at the (L-1)th future time point; fusing the multiple second decelerations corresponding to the vehicle at the (L-1)th future time point to obtain the fused second deceleration, including: fusing the multiple corrected second decelerations corresponding to the vehicle at the (L-1)th future time point to obtain the fused second deceleration.

[0013] In one possible implementation, based on the first deceleration of the target obstacle at the (L-1)th future time, the second deceleration of the vehicle at the (L-1)th future time is corrected to obtain the corrected second deceleration of the vehicle at the (L-1)th future time. This includes: determining a target time distance based on the planned trajectory point of the vehicle at the (L-1)th future time and the predicted trajectory point of the target obstacle at the (L-1)th future time. The target time distance represents the distance from the planned trajectory point of the vehicle at the (L-1)th future time to the predicted trajectory of the target obstacle at the (L-1)th future time. The time required for the target obstacle to reach the target time distance is used to adjust the first deceleration of the target obstacle at the (L-1)th future time, resulting in the adjusted first deceleration of the target obstacle at the (L-1)th future time. The adjusted first deceleration is not greater than the original first deceleration, and the adjustment range of the target time distance and the first deceleration is positively correlated. Based on the adjusted first deceleration of the target obstacle at the (L-1)th future time, the second deceleration of the vehicle at the (L-1)th future time is corrected, resulting in the corrected second deceleration of the vehicle at the (L-1)th future time.

[0014] In one possible implementation, selecting one of multiple planned trajectory information as the target planned trajectory information for the vehicle includes: determining the total cost corresponding to each planned trajectory information among the multiple planned trajectory information, wherein the total cost corresponding to the planned trajectory information is positively correlated with the cost corresponding to each future time, and for any future time, the cost corresponding to the future time is determined based on the distance between the planned trajectory point corresponding to the future time and the boundary indicated by the boundary information corresponding to the future time, which is used to represent the collision risk between the planned trajectory point corresponding to the future time and the boundary indicated by the boundary information corresponding to the future time; and selecting the planned trajectory information with the minimum corresponding total cost as the target planned trajectory information for the vehicle.

[0015] In one possible implementation, the boundary information includes a first boundary and a second boundary, wherein the distance between the first boundary and the current position of the vehicle is less than the distance between the second boundary and the current position of the vehicle. The method further includes: for any future time, determining a first-generation value based on the distance between the planned trajectory point corresponding to the future time and the first boundary corresponding to the future time, and determining a second-generation value based on the distance between the planned trajectory point corresponding to the future time and the second boundary corresponding to the future time; obtaining the weights corresponding to the first-generation value and the second-generation value, wherein the weight corresponding to the first-generation value is less than the weight corresponding to the second-generation value; and for any future time, performing a weighted calculation based on the first-generation value, the weight corresponding to the first-generation value, the second-generation value, and the weight corresponding to the second-generation value to determine the generation value corresponding to the future time.

[0016] Secondly, embodiments of this application propose a trajectory planning device, comprising: an acquisition module, configured to acquire multiple predicted trajectory information corresponding to each target obstacle, wherein each predicted trajectory information includes multiple predicted trajectory points corresponding to each future time, and the target obstacle includes obstacles within a preset range of the vehicle; a boundary determination module, configured to determine multiple boundary information corresponding to each future time of the target obstacle based on the multiple predicted trajectory information corresponding to the target obstacle, wherein the multiple boundary information corresponding to each future time is determined separately based on the multiple predicted trajectory points corresponding to each future time; and a trajectory planning module, configured to determine multiple planned trajectory information of the vehicle based on the multiple boundary information corresponding to each future time of the target obstacle, wherein each planned trajectory information includes multiple planned trajectory points corresponding to each future time; and to select one of the multiple planned trajectory information as the target planned trajectory information of the vehicle.

[0017] Thirdly, embodiments of this application provide an electronic device, including a processor and a memory, wherein: the memory is used to store computer programs; and the processor is used to execute the programs stored in the memory to implement the above-described method.

[0018] Fourthly, this application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the above-described method. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating a trajectory planning method provided in one embodiment of this application.

[0020] Figure 2 This is a schematic diagram of multi-track boundary generation for an obstacle vehicle provided in an embodiment of this application.

[0021] Figure 3 This is a flowchart illustrating a trajectory planning method provided in another embodiment of this application.

[0022] Figure 4 This is a schematic diagram of a deceleration limit generation process provided in an embodiment of this application.

[0023] Figure 5 This is a schematic diagram of the structure of a trajectory planning device provided in an embodiment of this application.

[0024] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0025] To make the technical problems, technical solutions, and beneficial effects solved by this application clearer, the following detailed description is provided in conjunction with embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0026] In related technologies, the trajectory information of obstacles is predicted, and then the trajectory information of the vehicle is planned by combining the predicted trajectory information of obstacles.

[0027] However, long-term research has revealed that the trajectory information of the planned vehicle in related technologies is not accurate enough, which can lead to abrupt changes in the planned response behavior. Specifically, in related technologies, when different predicted trajectories appear for obstacle vehicles, the trajectory with the highest probability is selected and planned based on this trajectory. However, changes in the predicted probability often lead to abrupt changes in trajectory selection, and the differences between different trajectories are significant, which in turn leads to inaccurate trajectory information planning for the vehicle, often resulting in abrupt changes in the planned response behavior.

[0028] Furthermore, in related technologies, the deceleration limit for following vehicles does not take into account multiple trajectories of the same obstacle vehicle, which often leads to excessive or insufficient deceleration. This results in poor comfort and safety in sudden change scenarios. Deceleration limits generally only consider the speed difference and position with the vehicle in front, without referencing the deceleration of the vehicle in front.

[0029] Furthermore, in related technologies, the deceleration limit range is not dynamically updated during the speed planning process. The deceleration limit is only calculated based on the current speed difference and position information. During the planning process, when the speed and position of the vehicle and the obstacle vehicle change, the existing deceleration limit often cannot plan a more reasonable speed trajectory that meets safety and comfort requirements.

[0030] Furthermore, in related technologies, velocity planning does not limit the deceleration range, has a large planning index range, and has low planning efficiency.

[0031] Furthermore, in related technologies, the handling of vehicles cutting in often determines whether the vehicle needs to decelerate based on the ideal following distance. Decelerating within the ideal following distance and not needing to decelerate beyond the ideal following distance often results in a lag where the vehicle is still decelerating due to distance limitations when the vehicle in front accelerates away.

[0032] In view of this, embodiments of this application propose a trajectory planning method, apparatus, electronic device, and storage medium, which can improve the accuracy of trajectory planning and reduce abrupt changes in planning response behavior.

[0033] Please see Figure 1 , Figure 1 This is a flowchart illustrating a trajectory planning method provided in one embodiment of this application. Figure 1The method shown can be applied to electronic devices, such as terminals or servers. Terminals can be smartphones, tablets, laptops, desktop computers, smart TVs, smart home devices, in-vehicle terminals (e.g., vehicle infotainment systems), etc., without specific limitations. Servers can be independent physical servers, server clusters or distributed systems composed of multiple physical servers, or cloud servers providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms. Figure 1 The methods shown may include: S110. Obtain multiple predicted trajectory information corresponding to each target obstacle. Each predicted trajectory information includes multiple predicted trajectory points corresponding to each future time. The target obstacles include obstacles within the preset range of the vehicle.

[0034] The predicted trajectory information can be derived using historical trajectory information corresponding to the target obstacle. For example, historical trajectory information of the target obstacle can be collected by sensors installed on the vehicle, and then multiple predicted trajectories corresponding to the target obstacle can be predicted using this historical trajectory information. The preset range can be related to the collection range capability of the sensors installed on the vehicle. In this embodiment, the target obstacle can be one or more. "Multiple" means no less than two. The target obstacle can include, but is not limited to, moving objects such as obstacle vehicles and pedestrians.

[0035] S120. Based on multiple predicted trajectory information corresponding to the target obstacle, determine multiple boundary information corresponding to the target obstacle at each future time. The multiple boundary information corresponding to the future time is determined based on the multiple predicted trajectory points corresponding to the future time.

[0036] Boundary information is used to indicate the boundaries formed by the target obstacle at future moments. In this embodiment, the boundary may include, but is not limited to, at least one of soft boundaries and hard boundaries. A soft boundary may refer to a boundary that the vehicle can touch during trajectory planning, but touching or even breaking through a soft boundary requires a certain cost. A hard boundary may refer to a boundary that restricts the vehicle from touching. In this embodiment, the distance between the hard boundary and the vehicle is greater than the distance between the soft boundary and the vehicle.

[0037] In this embodiment, for each of the multiple predicted trajectory points corresponding to the same future time, corresponding boundary information can be determined. This boundary information can cover the predicted trajectory point, and the extension direction of the boundary is perpendicular to the vehicle's travel direction.

[0038] S130. Based on multiple boundary information corresponding to the target obstacle at each future time, determine multiple planned trajectory information of the vehicle. Each planned trajectory information includes multiple planned trajectory points corresponding to each future time.

[0039] In this embodiment, multiple planned trajectory information of the vehicle can be determined based on the constraints of multiple boundary information corresponding to the target obstacle at each future time.

[0040] S140. Select one of the multiple planned trajectory information as the target planned trajectory information of the vehicle.

[0041] In this embodiment, after determining the target trajectory information of the vehicle, the trajectory of the vehicle can be planned according to the planning trajectory points corresponding to multiple future times in the target trajectory information, so that the vehicle arrives at the planning trajectory points indicated by the target trajectory information at each future time.

[0042] In this embodiment, multiple predicted trajectory information corresponding to each target obstacle is acquired. Each predicted trajectory information includes multiple predicted trajectory points corresponding to each future time. The target obstacle includes obstacles within a preset range of the vehicle. Based on the multiple predicted trajectory information corresponding to the target obstacle, multiple boundary information corresponding to the target obstacle at each future time is determined. The multiple boundary information corresponding to each future time is determined based on the multiple predicted trajectory points corresponding to each future time. Based on the multiple boundary information corresponding to the target obstacle at each future time, multiple planned trajectory information of the vehicle is determined. Each planned trajectory information includes multiple planned trajectory points corresponding to each future time. One of the multiple planned trajectory information is selected as the target planned trajectory information of the vehicle. In this way, the trajectory planning of the vehicle can be performed by combining multiple predicted trajectory information of the target obstacle. Since the trajectory planning is performed by combining multiple predicted trajectory information, more information can be referenced for predicting the trajectory, thereby improving the accuracy of the vehicle's trajectory planning.

[0043] In one possible implementation, multiple predicted trajectory information corresponding to each target obstacle is obtained, including: The vehicle acquires predicted trajectory information for at least one obstacle within a preset range. Based on the predicted trajectory information for each obstacle, the vehicle identifies a target obstacle and acquires multiple predicted trajectory information for each target obstacle. One of the trajectory points in one of the predicted trajectory information for the target obstacle is located in the vehicle's direction of travel.

[0044] In this embodiment, one of the trajectory points in the corresponding predicted trajectory information that is an obstacle in the direction of the vehicle's travel is taken as the target obstacle, while other obstacles are taken as non-target obstacles and are not considered. This allows for the filtering of obstacles, thereby reducing computing resources.

[0045] In another possible implementation, all obstacles could be treated as target obstacles.

[0046] In one possible implementation, the boundary information includes a first boundary and a second boundary, wherein the distance between the first boundary and the vehicle's current position is less than the distance between the second boundary and the vehicle's current position. Based on multiple predicted trajectory information corresponding to the target obstacle, multiple boundary information corresponding to the target obstacle at each future time is determined, including: Based on multiple predicted trajectory information corresponding to the target obstacle, a spatiotemporal map is generated. The spatiotemporal map is used to represent multiple occupied areas corresponding to the target obstacle at multiple future times. Each occupied area is determined based on a predicted trajectory point in one of the predicted trajectory information. The occupied area is located in the direction of travel of the vehicle. For any future time, a second boundary is determined based on the overlapping part between the multiple occupied areas corresponding to the future time, and a first boundary is determined based on the non-overlapping part corresponding to the future time. The non-overlapping part includes the part of the total area formed by the multiple occupied areas other than the overlapping part.

[0047] The Space-Time (ST) diagram is a tool used to visualize and analyze the distribution and changes of vehicle motion states (including both the vehicle and obstacles) in space (usually longitudinal position, such as the length of the road) and time, providing a basis for decision-making such as trajectory planning. The horizontal axis of the ST diagram typically represents space, generally the vehicle's longitudinal position on the road (e.g., distance from a reference point, in meters); the vertical axis represents time, in seconds. In the ST diagram, each curve or region can represent a vehicle's trajectory or the space-time range it occupies. The first boundary can be understood as a soft boundary, i.e., the boundary the vehicle can touch, and the second boundary can be understood as a hard boundary, i.e., the boundary the vehicle cannot touch. In this embodiment, the occupied area can be a region determined by combining the predicted trajectory point and the size of the target obstacle. For example, the predicted trajectory point can be the center of the target obstacle, and the region can be determined by combining the length and width of the target obstacle. The length can be along the vehicle's direction of travel, and the width can be along the perpendicular direction to the vehicle's direction of travel; there are no restrictions here.

[0048] Please see Figure 2 , Figure 2This is a schematic diagram of multi-track boundary generation for an obstacle vehicle provided in an embodiment of this application.

[0049] like Figure 2 As shown in (a), EGO represents the autonomous vehicle (autonomous vehicle), located on the left side of the road, with a tendency to move upwards. OBS represents the obstacle vehicle (one type of obstacle), and the dashed line indicates the predicted trajectory information of the obstacle vehicle. P1, P2, and P3 are three different predicted trajectory information, that is, different positions that the obstacle vehicle may travel to (corresponding to different trajectory selections).

[0050] Will Figure 2 (a) in the text can be mapped to, for example, Figure 2 The ST diagram shown in (b) is an example. The horizontal axis of the ST diagram represents time (T), and the vertical axis represents space (S, usually referring to the longitudinal position, such as the length of a road, which can be understood as the direction of the vehicle's travel). It is used to combine "spatial position" with "time" to analyze the dynamic occupancy range of a vehicle. 1. Spatiotemporal distribution of obstacle areas: The tilted "block-shaped areas" in the obstacle vehicle (OBS) diagram represent the occupancy range (also known as the occupied area) of the obstacle vehicle in different time intervals: the T-axis represents the time span of the obstacle vehicle occupying the space; the S-axis of the area represents the spatial coverage range of the obstacle vehicle; the tilt of the area reflects the speed of the obstacle vehicle (the steeper the tilt, the faster the speed). Division of hard and soft boundaries: soft boundary: a soft boundary is marked "non-overlapping is a soft boundary", which refers to the occupied area of ​​a single obstacle vehicle (or one that does not overlap with other obstacle areas). Soft boundaries are areas that need to be treated with caution, but are not completely prohibited. There is a certain risk when a vehicle enters, but the risk can be quantified by a cost function (such as the soft boundary cost formula mentioned earlier). Hard Boundaries: Marked as "overlapping areas are hard boundaries," these refer to the overlapping areas occupied by different obstacle vehicles (or different trajectories of the same obstacle vehicle). Hard boundaries are absolutely inviolable areas. If the planned trajectory of the vehicle enters this area, it will be judged as having an extremely high risk of collision (or even an inevitable collision), and must be strictly avoided during trajectory planning. In summary, P1, P2, and P3 are three predicted entry trajectories that enter the vehicle's lane at different times. After trajectory collision detection, the ST diagram on the right is generated. Trajectory P3, with its overlapping area, is a hard boundary. P1 / P2, excluding the overlapping area, is a soft boundary. When calculating deceleration, the safety distance is not considered, and the intersection of the planned point with this part is not considered during the planning process.

[0051] In this embodiment, a spatiotemporal map is generated based on multiple predicted trajectory information corresponding to the target obstacle. The spatiotemporal map is used to represent multiple occupied areas corresponding to the target obstacle at multiple future times. Each occupied area is determined based on a predicted trajectory point in one of the predicted trajectory information. The occupied area is located in the direction of travel of the vehicle. For any future time, a second boundary is determined based on the overlapping part between the multiple occupied areas corresponding to the future time, and a first boundary is determined based on the non-overlapping part corresponding to the future time. The non-overlapping part includes the part of the total area formed by the multiple occupied areas other than the overlapping part. Since the planning of the trajectory information is combined with the first boundary and the second boundary, the accuracy of the vehicle's trajectory planning can be improved.

[0052] In another possible implementation, the trajectory information can be determined through the first or second boundary, which can reduce computing resources.

[0053] In one possible implementation, multiple planned trajectory information of the vehicle is determined based on multiple boundary information corresponding to the target obstacle at various future time points, including: For the first future moment among multiple future moments, based on the vehicle's current position and the multiple boundary information corresponding to the target obstacle at the first future moment, at least one planned trajectory point corresponding to the first future moment is determined; for the Lth future moment among multiple future moments, based on the planned trajectory point corresponding to the vehicle at the (L-1)th future moment and the multiple boundary information corresponding to the target obstacle at the Lth future moment, at least one planned trajectory point corresponding to the Lth future moment is determined, where L is an integer not less than 2; based on the planned trajectory points corresponding to each of the multiple future moments, multiple planned trajectory information of the vehicle is determined.

[0054] In this embodiment, at least one planned trajectory point corresponding to the first future time can be determined based on the constraints of the vehicle's current position and multiple boundary information corresponding to the target obstacle at the first future time. For the Lth future time, at least one planned trajectory point corresponding to the Lth future time is determined based on the constraints of the planned trajectory point of the vehicle at the (L-1)th future time and the multiple boundary information corresponding to the target obstacle at the Lth future time. This allows for multiple sets of planned trajectory point groups, each group comprising consecutive planned trajectory points corresponding to multiple future times. In other words, the planned trajectory point corresponding to the (L-1)th future time serves as the starting point for the planned trajectory point corresponding to the Lth future time. Each set of planned trajectory point groups can be considered as one planned trajectory information. The Lth future time is later than the (L-1)th future time.

[0055] In this embodiment, for the first future moment among multiple future moments, based on the vehicle's current position and multiple boundary information corresponding to the target obstacle at the first future moment, at least one planned trajectory point corresponding to the first future moment is determined. For the Lth future moment among multiple future moments, based on the planned trajectory point corresponding to the vehicle at the (L-1)th future moment and the multiple boundary information corresponding to the target obstacle at the Lth future moment, at least one planned trajectory point corresponding to the Lth future moment is determined, where L is an integer not less than 2. Based on the planned trajectory points corresponding to each of the multiple future moments, multiple planned trajectory information of the vehicle is determined. That is, the planned trajectory point corresponding to each future moment is determined by the planned trajectory point corresponding to the previous future moment. This makes the multiple planned trajectory points in the obtained multiple planned trajectory information more coherent, thereby improving the coherence of the vehicle's trajectory planning.

[0056] In one possible implementation, the predicted trajectory information also includes a first velocity corresponding to each of multiple future time moments. Based on the planned trajectory point of the vehicle at the (L-1)th future time moment and multiple boundary information of the target obstacle at the Lth future time moment, at least one planned trajectory point corresponding to the Lth future time moment is determined, including: Based on the planned trajectory point of the vehicle at the (L-1)th future time, the second speed of the vehicle at the (L-1)th future time, multiple boundary information of the target obstacle at the Lth future time, and multiple first speeds of the target obstacle at the L-1th future time, multiple second decelerations of the vehicle at the L-1th future time are determined, with each of the multiple boundary information, multiple first speeds, and multiple second decelerations corresponding one-to-one. The multiple second decelerations of the vehicle at the L-1th future time are fused to obtain the fused second deceleration. Based on the fused second deceleration of the vehicle at the L-1th future time, the safe area of ​​the vehicle at the Lth future time is determined. Based on the points within the safe area of ​​the vehicle at the Lth future time, at least one planned trajectory point at the Lth future time is determined.

[0057] In this embodiment, for the first future time, based on the first velocity corresponding to the current time (the time corresponding to the current position), the current position, multiple boundary information of the target obstacle at the first future time, and multiple first velocities of the target obstacle at the current time, multiple second decelerations corresponding to the vehicle at the (L-1)th future time are determined. Specifically, for each predicted trajectory information, based on the multiple boundary information corresponding to the Lth future time, the multiple first velocities of the target obstacle at the L-1th future time, the planned trajectory point of the vehicle at the L-1th future time, and the second velocity of the vehicle at the L-1th future time, the second deceleration of the vehicle at the L-1th future time can be determined. Therefore, the vehicle can decelerate at this second deceleration during the time between the L-1th and Lth future times, thus obtaining multiple second decelerations corresponding to multiple predicted trajectory information. Then, the second decelerations corresponding to the multiple predicted trajectory information can be fused to obtain the fused deceleration. Then, using the fused second deceleration, the safe zone corresponding to the vehicle at the Lth future time is determined, and the points within the safe zone corresponding to the vehicle at the Lth future time are used as the planned trajectory points at the Lth future time. Specifically, the safe zone corresponding to the Lth future time is the area between the displacement determined based on the fused second deceleration and second velocity at the (L-1)th future time and the planned trajectory point at the (L-1)th future time.

[0058] It should be noted that if the second speed corresponding to L-1 future moments is lower than the first speed corresponding to L-1 future moments, then if both the obstacle and the vehicle move at a constant speed between the L-1 future moment and the Lth future moment, the distance between the vehicle and the obstacle will increase. Conversely, if the second speed corresponding to L-1 future moments is higher than the first speed corresponding to L-1 future moments, the distance between the vehicle and the obstacle will decrease. Therefore, a second deceleration can be set to prevent the vehicle from colliding with the obstacle when the Lth future moment is reached.

[0059] In this embodiment, multiple predicted trajectory information are generated, and each predicted trajectory information can determine a corresponding second deceleration. These multiple second decelerations can then be fused. It should be noted that the fusion method can be, for example, averaging.

[0060] In this embodiment, based on the planned trajectory point of the vehicle at the (L-1)th future time, the second speed of the vehicle at the (L-1)th future time, multiple boundary information of the target obstacle at the Lth future time, and multiple first speeds of the target obstacle at the L-1th future time, multiple second decelerations of the vehicle at the L-1th future time are determined, with each of the multiple boundary information, multiple first speeds, and multiple second decelerations corresponding one-to-one. The multiple second decelerations of the vehicle at the L-1th future time are fused to obtain a fused second deceleration. Based on the fused second deceleration of the vehicle at the L-1th future time, a safe area of ​​the vehicle at the Lth future time is determined. Based on the points within the safe area of ​​the vehicle at the Lth future time, at least one planned trajectory point at the Lth future time is determined. Thus, since the planned trajectory point is determined by fusing multiple second decelerations corresponding to future times, the accuracy of the determined planned trajectory point can be improved.

[0061] Another possible implementation is to disregard the speed of obstacles, which would reduce the required computing resources.

[0062] In one possible implementation, multiple second decelerations corresponding to the vehicle at the (L-1)th future time are fused to obtain a fused second deceleration, including: Obtain the weight corresponding to each second deceleration. The weight corresponding to the second deceleration is related to the prediction probability corresponding to the predicted trajectory information of the first velocity corresponding to the second deceleration. Based on the weight corresponding to each second deceleration, perform weighted calculation on the multiple second decelerations corresponding to the vehicle at the (L-1)th future time to obtain the fused second deceleration.

[0063] In this embodiment, the prediction probability corresponding to the predicted trajectory information of the first velocity corresponding to the second deceleration can be used as the fusion weight corresponding to the second deceleration.

[0064] For example, assuming there is predicted trajectory information 1 and planned trajectory information 2, with the probability corresponding to predicted trajectory information 1 being P1 and the probability corresponding to planned trajectory information 2 being P2, then based on the predicted trajectory points and the first velocity corresponding to each future time in predicted trajectory information 1, the second deceleration 1 corresponding to each future time can be determined. Based on the predicted trajectory points and the first velocity corresponding to each future time in predicted trajectory information 2, the second deceleration 2 corresponding to each future time can be determined. Therefore, when fusing the second deceleration, it can be the second deceleration 1*P1 + the second deceleration 2*P2.

[0065] In this embodiment, by obtaining the weights corresponding to each second deceleration, the weights corresponding to the second decelerations are related to the prediction probability corresponding to the predicted trajectory information of the first velocity corresponding to the second deceleration. Based on the weights corresponding to each second deceleration, the multiple second decelerations corresponding to the vehicle at the (L-1)th future time are weighted and calculated to obtain the fused second deceleration. This allows for the fusion of the second decelerations by combining the probability of the predicted trajectory information, thereby improving the accuracy of the determined second decelerations. This improves the accuracy of the determined safe area, and further improves the accuracy of the determined second prediction estimation point, which is beneficial to improving the accuracy of the vehicle trajectory planning.

[0066] In one possible implementation, the predicted trajectory information also includes first decelerations corresponding to multiple future time points. Before fusing the multiple second decelerations corresponding to the vehicle at the (L-1)th future time point to obtain the fused second deceleration, the method further includes: Obtain the first deceleration of the target obstacle at the (L-1)th future time; based on the first deceleration of the target obstacle at the (L-1)th future time, correct the second deceleration of the vehicle at the (L-1)th future time to obtain the corrected second deceleration of the vehicle at the (L-1)th future time.

[0067] Accordingly, the multiple second decelerations corresponding to the vehicle at the (L-1)th future time are fused to obtain the fused second deceleration, including: fusing the multiple corrected second decelerations corresponding to the vehicle at the (L-1)th future time to obtain the fused second deceleration.

[0068] In this embodiment, the correction method can be to add the absolute value of the first deceleration to the absolute value of the second deceleration to obtain the corrected absolute value of the second deceleration.

[0069] In this embodiment, the first deceleration of the target obstacle at the (L-1)th future time is obtained; based on the first deceleration of the target obstacle at the (L-1)th future time, the second deceleration of the vehicle at the (L-1)th future time is corrected to obtain the corrected second deceleration of the vehicle at the (L-1)th future time. The multiple corrected second decelerations of the vehicle at the (L-1)th future time are fused to obtain the fused second deceleration. Therefore, after determining the corresponding second deceleration based on the predicted trajectory information, the first deceleration corresponding to the predicted trajectory information can also be used to correct the second deceleration, which improves the safety of the vehicle's trajectory planning.

[0070] In one possible implementation, the second deceleration obtained from the lookup table can be determined. Based on the second deceleration obtained from the lookup table and the fused second deceleration, a second deceleration for calculating the safe region is determined. For example, the smaller value between the second deceleration obtained from the lookup table and the fused second deceleration can be selected, or the average value of the second deceleration obtained from the lookup table and the fused second deceleration can be selected as the second deceleration for calculating the safe region. No limitation is imposed here. Optionally, the table can record the velocity difference between the second velocity and the first velocity, as well as the mapping relationship between the second decelerations.

[0071] Table 1 shows the interpolation table between velocity difference and deceleration.

[0072] Table 1 Linear interpolation (denoted as a function) The deceleration limit is obtained, and then the smaller of this limit and the reference deceleration of the obstacle vehicle is taken to obtain the final maximum deceleration limit. .

[0073] in, This represents the second deceleration obtained from the table lookup. This represents the second deceleration rate of the fusion.

[0074] For example, when the vehicle is faster than the obstacle vehicle in front, the two vehicles gradually close in. When the speed difference is large, a safe deceleration needs to be calculated based on the speed difference and positional relationship. The calculation principle is that when the vehicle reaches its maximum deceleration in the next second, it can decelerate to the same speed as the vehicle in front when it reaches the set safe distance, based on the basic formula. After substituting the relevant distance and velocity parameters, we obtain: ; ; ; ; This indicates the start time when the obstacle vehicle is projected onto the ST map after completing the collision detection, whichever is greater than or equal to 1 second, but at least 1 second. This indicates the safe distance; for soft boundary obstacles, this value is 0 (the vehicle can collide with the soft boundary). This indicates the distance deviation caused by the system's execution delay. The acceleration is the initial acceleration of the vehicle. In this embodiment, when the boundary information includes a first boundary and a second boundary, a safe boundary is determined based on the safe distance and the second boundary. The distance between the safe boundary and the first boundary is less than the distance between the second boundary and the first boundary. Then, the boundary closer to the vehicle, either the first boundary or the safe boundary, is selected to calculate the corresponding second deceleration.

[0075] For obstacles with multiple predicted trajectories, the deceleration calculation is based on the probability weighting of different trajectories to obtain the final deceleration value, which is then compared with the deceleration obtained from the lookup table and the smaller value is taken.

[0076] ; Where k represents the number of predicted trajectory information, k=0 represents the second deceleration determined based on the first predicted trajectory information, and k=n represents the second deceleration determined based on the last predicted trajectory information.

[0077] In one possible implementation, based on the first deceleration of the target obstacle at the (L-1)th future time, the second deceleration of the vehicle at the (L-1)th future time is corrected to obtain the corrected second deceleration of the vehicle at the (L-1)th future time, including: Based on the planned trajectory point of the vehicle at the (L-1)th future time and the predicted trajectory point of the target obstacle at the (L-1)th future time, a target time distance is determined. The target time distance represents the time required for the vehicle to travel from the planned trajectory point at the (L-1)th future time to the predicted trajectory point of the target obstacle at the (L-1)th future time. Based on the target time distance, the first deceleration of the target obstacle at the (L-1)th future time is adjusted to obtain the adjusted first deceleration of the target obstacle at the (L-1)th future time. The adjusted first deceleration is not greater than the original first deceleration, and the target time distance is positively correlated with the adjustment range of the first deceleration. Based on the adjusted first deceleration of the target obstacle at the (L-1)th future time, the second deceleration of the vehicle at the (L-1)th future time is corrected to obtain the corrected second deceleration of the vehicle at the (L-1)th future time.

[0078] In this embodiment, a smaller target time distance indicates a closer proximity between the vehicle and the obstacle, making a collision more likely. Therefore, the first deceleration corresponding to the obstacle needs to be considered. Conversely, a larger target time distance indicates a greater distance between the vehicle and the obstacle, making a collision less likely. In this case, the focus on the first deceleration corresponding to the obstacle can be reduced. The target time distance can be determined by combining the second velocity and second deceleration corresponding to the vehicle at the (L-1)th future time.

[0079] For example, the deceleration of the obstacle vehicle is determined based on the distance, and a table of time-distance and deceleration ratios is established. The closer the distance, the higher the deceleration attenuation factor (denoted as...). The larger the distance, the smaller the deceleration attenuation factor of the obstacle vehicle; the corresponding table of time distance and deceleration attenuation factor used in this method is shown in Table 2: Table 2 The relative distance between the obstacle vehicle and the vehicle is within 2 seconds, during which the deceleration does not decrease, and decreases to 0 at 4 seconds. , where S is the relative position of the obstacle vehicle and the vehicle, and V is the speed of the vehicle.

[0080] In this embodiment, a target time distance is determined based on the planned trajectory point of the vehicle at the (L-1)th future time and the predicted trajectory point of the target obstacle at the (L-1)th future time. The target time distance represents the time required for the vehicle to travel from the planned trajectory point at the (L-1)th future time to the predicted trajectory point of the target obstacle at the (L-1)th future time. Based on the target time distance, the first deceleration of the target obstacle at the (L-1)th future time is adjusted to obtain the adjusted first deceleration of the target obstacle at the (L-1)th future time. The adjusted first deceleration is not greater than the original first deceleration, and the target time distance is positively correlated with the adjustment range of the first deceleration. Based on the adjusted first deceleration of the target obstacle at the (L-1)th future time, the second deceleration of the vehicle at the (L-1)th future time is corrected to obtain the corrected second deceleration of the vehicle at the (L-1)th future time. This approach balances the safety of the vehicle's trajectory planning with the comfort of the ride.

[0081] It should be noted that after determining the target trajectory information, the planned trajectory points, second velocity, and second deceleration corresponding to each of the multiple future moments in the target trajectory information can be smoothed to obtain the smoothed planned trajectory points and second velocity. Thus, the smoothed planned trajectory points and second velocity can be used as the motion of the vehicle at multiple future moments.

[0082] The following explains how to determine the target planning trajectory information.

[0083] In one possible implementation, one of multiple planned trajectory information is selected as the target planned trajectory information for the vehicle, including: The total cost value corresponding to each of the multiple planned trajectory information is determined. The total cost value corresponding to the planned trajectory information is positively correlated with the cost value corresponding to each future time. For any future time, the cost value corresponding to the future time is determined based on the distance between the planned trajectory point corresponding to the future time and the boundary indicated by the boundary information corresponding to the future time, and is used to represent the collision risk between the planned trajectory point corresponding to the future time and the boundary indicated by the boundary information corresponding to the future time. The planned trajectory information with the minimum corresponding total cost value is taken as the target planned trajectory information of the vehicle.

[0084] In this embodiment, the smaller the distance between the planned trajectory point and the boundary indicated by the boundary information corresponding to the future time, the greater the cost, which means the higher the collision risk.

[0085] In this embodiment, the total cost value corresponding to each planned trajectory information among multiple planned trajectory information is determined. The total cost value corresponding to the planned trajectory information is positively correlated with the cost value corresponding to each future time. For any future time, the cost value corresponding to the future time is determined based on the distance between the planned trajectory point corresponding to the future time and the boundary indicated by the boundary information corresponding to the future time, which is used to represent the collision risk between the planned trajectory point corresponding to the future time and the boundary indicated by the boundary information corresponding to the future time. The planned trajectory information with the smallest corresponding total cost value is taken as the target planned trajectory information of the vehicle. In this way, selecting the minimum total cost value is also selecting the target planned trajectory information with the lowest collision risk, thereby improving the safety of the vehicle's trajectory planning.

[0086] In one possible implementation, the boundary information includes a first boundary and a second boundary, wherein the distance between the first boundary and the vehicle's current position is less than the distance between the second boundary and the vehicle's current position. The method further includes: For any future moment, the first-generation value is determined based on the distance between the planned trajectory point corresponding to the future moment and the first boundary corresponding to the future moment, and the second-generation value is determined based on the distance between the planned trajectory point corresponding to the future moment and the second boundary corresponding to the future moment; the weights corresponding to the first-generation value and the second-generation value are obtained, and the weight corresponding to the first-generation value is less than the weight corresponding to the second-generation value; for any future moment, the generation value corresponding to the future moment is determined by weighted calculation based on the first-generation value, the weight corresponding to the first-generation value, the second-generation value, and the weight corresponding to the second-generation value.

[0087] In this embodiment, the cost mapping functions for the first boundary and the second boundary are different. Specifically, when the distance to the first boundary and the distance to the second boundary are the same, the cost corresponding to the second boundary is greater than the cost corresponding to the first boundary. In other words, the cost of being closer to the second boundary is greater than the cost of being closer to the first boundary; or, the collision risk of being closer to the second boundary is greater than the collision risk of being closer to the first boundary.

[0088] In this embodiment, for any future moment, a first-generation value is determined based on the distance between the planned trajectory point corresponding to the future moment and the first boundary corresponding to the future moment, and a second-generation value is determined based on the distance between the planned trajectory point corresponding to the future moment and the second boundary corresponding to the future moment. The weights corresponding to the first-generation value and the second-generation value are obtained, with the weight of the first-generation value being less than the weight of the second-generation value. For any future moment, the generation value corresponding to the future moment is determined by weighted calculation based on the first-generation value, the weight of the first-generation value, the second-generation value, and the weight of the second-generation value. This allows for a more accurate assessment of collision risk, thereby improving the safety of the vehicle's trajectory planning.

[0089] For example, the cost calculation for obstacle vehicles involves generating soft and hard boundaries on the ST map based on the different trajectories of the obstacle vehicles. During the planning process, the obstacle vehicle cost at a planning point is obtained according to the location of the planning point and the distances to the soft and hard boundaries using the following formula. When calculating the obstacle cost at a planning point, the greater the intrusion distance, the higher the cost; the cost is 0 if the intrusion distance is greater than the intrusion distance. The obstacle cost for soft boundaries does not consider obstacle intersections; the calculation method is as follows: ; ; in, For hard boundary barrier weights, Soft boundary barrier weights, where , For ideal following distance, The lower boundary of the obstacle. The location of the planning point (also known as the planning trajectory point).

[0090] For ease of understanding, the following examples, based on any of the above examples, illustrate how to plan the trajectory of the vehicle.

[0091] Please see Figure 3 , Figure 3 A flowchart illustrating a trajectory planning method provided in another embodiment of this application is shown below. Figure 3 The methods shown may include: S310, Start planning.

[0092] S320. Filter out the target obstacle vehicle based on the trajectory of the obstacle vehicle, and take all the predicted trajectories of the filtered target obstacle vehicle.

[0093] In this embodiment, the trajectory of the obstacle is predicted, and the description of the predicted trajectory information can be referred to.

[0094] S330: Traverse all target obstacle vehicles and perform collision detection with the vehicle's trajectory. Project the collision detection onto the ST map to obtain the obstacle vehicle boundaries. Merge different boundaries of the same target obstacle vehicle. The overlapping part is used as the hard boundary, and the non-overlapping part is used as the soft boundary.

[0095] The ST chart can be referenced. Figure 2 Explanation.

[0096] S340, Start longitudinal velocity dynamic programming.

[0097] S350. Calculate the deceleration limit of the planning point at the current moment. Based on the velocity and acceleration information of the planning point at the current moment, traverse the hard and soft boundaries of all target obstacle vehicles, calculate the deceleration limit of each boundary, merge the deceleration limits of the same target obstacle vehicle, and take the minimum value of the deceleration limits of all target obstacle vehicles.

[0098] The deceleration limit can be referenced from the description of the second deceleration, and is not limited here. The planning point is also called the planning trajectory point. The current time can be referenced from the description of the (L-1)th future time. In this embodiment, the smaller the deceleration limit, the larger the absolute value of the deceleration limit.

[0099] S360. Based on the deceleration and acceleration limits, obtain the planning point for the next time step, calculate the total obstacle cost for each planning point, and update the deceleration limit of the planning point based on the planning point information.

[0100] The next time point can be referred to in the description of the Lth future time point.

[0101] S370. Determine whether the planned point has reached the end point.

[0102] The endpoint can be the planned trajectory point corresponding to the last future moment.

[0103] S380. Based on the planned trajectory with the minimum total cost, the deceleration limit of the starting point and the deceleration limit on the trajectory corresponding to the minimum total cost are combined to obtain the deceleration limit of this plan as the deceleration boundary of QP.

[0104] S390 and QP planning are used to smooth the speed trajectory.

[0105] S400, output speed trajectory planning.

[0106] In this embodiment, obstacle vehicles with collision risks are identified and screened based on the currently planned trajectory. All their predicted trajectories (including position, speed, and acceleration information) and probability information are collected. Then, all trajectories of each obstacle vehicle are traversed, and collision checks are performed on each trajectory and the vehicle to obtain the boundary values ​​in the ST coordinate system. Each obstacle vehicle will overlap with the boundary on the ST. The overlapping part is a hard boundary that cannot intersect with the planned trajectory of the vehicle, and the non-overlapping part is a soft boundary that can intersect with the planned trajectory, and dynamic planning begins. Each time a planning point is indexed to the next point, based on the position and velocity information of that planning point, as well as the soft and hard boundaries of the obstacle vehicles at the corresponding time, the deceleration limit from the current planning point to the next point is calculated. This deceleration limit is then used to select the next planning point. For all planning points at each time step, the boundaries of all relevant obstacle vehicles are determined based on the planning point's position information. The ideal following distances of these obstacle vehicles are traversed, and the obstacle cost for each planning point is calculated. If it's a soft boundary, collision is not considered (the planning point can be within the soft boundary); if it's a hard boundary, collision is considered (the planning point cannot be within the hard boundary). Then, it is determined whether the target time position has been reached. Otherwise, some steps are continued (e.g., for all planning points at each time step, the boundaries of all relevant obstacle vehicles are determined based on the planning point's position information, the ideal following distances of these obstacle vehicles are traversed, the obstacle cost for each planning point is calculated, and the target time position is determined), until all indexable planning points have completed cost calculation and deceleration limit assignment. Based on the costs of all endpoints, the discrete trajectory with the minimum cost is selected as the dynamic trajectory. The dynamic programming (DP) program outputs a speed planning trajectory and records the deceleration limits on this trajectory as the deceleration limits for this planning. This deceleration limit is then output to the quadratic programming (QP) program as the deceleration boundary for optimization. Based on the DP results, a higher-resolution discrete speed trajectory is obtained through interpolation. The minimum value of all deceleration limits calculated by DP is used as the deceleration boundary for QP, completing the QP program and obtaining a smooth, comfortable speed planning trajectory that conforms to dynamic following distance.

[0107] The following explains how to determine the deceleration limit.

[0108] Please see Figure 4 , Figure 4 This is a schematic diagram illustrating a deceleration limit generation process provided in an embodiment of this application. Figure 4 The methods shown may include: S410. Select target obstacle vehicles and record all predicted trajectories of the target obstacle vehicles.

[0109] S420. Generate ST diagram and generate different boundaries.

[0110] In this example, different boundaries can include soft boundaries and hard boundaries.

[0111] S430, Traversing the target obstacle vehicle.

[0112] S440, Determine if there are multiple predicted trajectories.

[0113] S450 generates deceleration limits based on the position, speed, and acceleration information of the target obstacle vehicle, and the position, speed, and acceleration information of the vehicle itself.

[0114] S460: Calculate the deceleration of each boundary based on the soft boundary, hard boundary, the position, speed, and acceleration information of the target obstacle vehicle, and the speed, position, and acceleration information of the vehicle itself, and then weight them to obtain the deceleration limit.

[0115] S470. Finally, the minimum deceleration limit among all the target obstacle vehicles is taken as the final deceleration limit.

[0116] In summary, this embodiment discloses an improved dynamic programming algorithm for autonomous driving based on the calculation of safe deceleration limits for multiple predicted trajectories of obstacle vehicles. It designs a speed planning algorithm that, while ensuring safety, better reflects actual following behavior. This algorithm generates safe deceleration limits, soft boundaries, and hard boundaries through multiple predicted trajectories and probabilities of obstacle vehicles. The dynamic speed planning algorithm is based on deceleration limits and different boundary cost calculation methods for obstacle vehicles. First, by receiving path information from lateral planning, obstacle vehicle information for longitudinal following is selected based on different predicted trajectories. Boundaries in the obstacle vehicle ST graph are generated based on collision detection. Deceleration limits are calculated based on different boundaries, and the planned safe deceleration limits are obtained by combining the decelerations obtained from multiple obstacles. During the dynamic planning process, the cost values ​​of different boundaries for different obstacle vehicles are dynamically adjusted based on the planned speed information. These cost values ​​are incorporated into the speed planning optimization problem, effectively improving the accuracy of responding to different obstacle behaviors. This algorithm can meet the needs of scenarios with uncertain obstacle behaviors, especially sudden changes such as emergency cut-ins and lane-jumping, achieving safer and more timely speed planning.

[0117] Compared to planning methods that only consider a single trajectory of the obstacle vehicle, this embodiment quantifies different predicted trajectories of the obstacle vehicle into the planning process by using deceleration limits and setting soft and hard boundaries. This allows for the comprehensive utilization of predicted probability information and different trajectory information, enabling better handling of sudden changes in predictions while ensuring safety and achieving smoother planning. Secondly, while previous methods handled the vehicle's acceleration and deceleration behavior based on following distance, this embodiment proposes a method to determine deceleration limits based on the actual speed, acceleration, and position information of both the vehicle and the obstacle vehicle. This reduces the dependence of the vehicle's behavior on following distance, improves following efficiency, and better aligns with human-centered driving styles. Finally, this embodiment incorporates dynamic deceleration limit calculations into the DP planning process, increasing the planning feasibility and meeting deceleration requirements under changing planning environments. This maintains the rationality of speed planning and improves the efficiency of longitudinal speed planning.

[0118] The technical effects of this embodiment are as follows: When following another vehicle, the deceleration limit calculation method based on the present invention, obtained according to the relative vehicle speed and position information, satisfies the actual scenario, which can weaken the influence of following distance, enable rapid response during scene switching, and improve following efficiency, especially when a fast vehicle cuts in, it can control the vehicle speed well; in addition, since the deceleration limit can greatly reduce the search scale of dynamic programming, it can improve the efficiency of speed planning; at the same time, when dealing with the variability of the predicted behavior of obstacle vehicles, it can take into account different predicted trajectories, and achieve the continuity of planning under the premise of ensuring safety, thereby improving the smoothness of driving control.

[0119] The method embodiments have been described above; the product embodiments are described below.

[0120] Please see Figure 5 , Figure 5 This is a schematic diagram of the structure of a trajectory planning device provided in one embodiment of this application. Figure 3 The illustrated device can be applied to electronic devices, and the device may include an acquisition module 510, a boundary determination module 520, and a trajectory planning module 530, wherein: The acquisition module 510 is used to acquire multiple predicted trajectory information corresponding to each target obstacle. Each predicted trajectory information includes multiple predicted trajectory points corresponding to each future time. The target obstacle includes obstacles within a preset range of the vehicle. The boundary determination module 520 is used to determine multiple boundary information corresponding to each future time of the target obstacle based on the multiple predicted trajectory information corresponding to the target obstacle. The multiple boundary information corresponding to each future time is determined separately based on the multiple predicted trajectory points corresponding to each future time. The trajectory planning module 530 is used to determine multiple planned trajectory information of the vehicle based on the multiple boundary information corresponding to each future time of the target obstacle. Each planned trajectory information includes multiple planned trajectory points corresponding to each future time. One of the multiple planned trajectory information is selected as the target planned trajectory information of the vehicle.

[0121] In one possible implementation, the boundary information includes a first boundary and a second boundary. The distance between the first boundary and the current position of the vehicle is less than the distance between the second boundary and the current position of the vehicle. When the boundary determination module 520 determines the multiple boundary information corresponding to the target obstacle at each future moment based on multiple predicted trajectory information corresponding to the target obstacle, it is used to: generate a spatiotemporal map based on the multiple predicted trajectory information corresponding to the target obstacle. The spatiotemporal map is used to represent multiple occupied areas corresponding to the target obstacle at each of the multiple future moments. Each occupied area is determined based on a predicted trajectory point in one of the predicted trajectory information. The occupied area is located in the direction of travel of the vehicle. For any future moment among the multiple future moments, a second boundary is determined based on the overlapping part between the multiple occupied areas corresponding to the future moment, and a first boundary is determined based on the non-overlapping part corresponding to the future moment. The non-overlapping part includes the part other than the overlapping part in the total area formed by the multiple occupied areas.

[0122] In one possible implementation, when the trajectory planning module 530 determines multiple planned trajectory information of the vehicle based on multiple boundary information corresponding to the target obstacle at each future time, it is used to: for the first future time among multiple future times, determine at least one planned trajectory point corresponding to the first future time based on the current position of the vehicle and the multiple boundary information corresponding to the target obstacle at the first future time; for the Lth future time among multiple future times, determine at least one planned trajectory point corresponding to the Lth future time based on the planned trajectory point corresponding to the vehicle at the (L-1)th future time and the multiple boundary information corresponding to the target obstacle at the Lth future time, where L is an integer not less than 2; and determine multiple planned trajectory information of the vehicle based on the planned trajectory points corresponding to each of the multiple future times.

[0123] In one possible implementation, the predicted trajectory information also includes multiple first velocities corresponding to each future time. When the trajectory planning module 530 determines at least one planned trajectory point corresponding to the Lth future time based on the planned trajectory point of the vehicle at the L-1th future time and multiple boundary information of the target obstacle at the Lth future time, it is used to: determine multiple second decelerations of the vehicle at the L-1th future time based on the planned trajectory point of the vehicle at the L-1th future time, the second velocity of the vehicle at the L-1th future time, the multiple boundary information of the target obstacle at the Lth future time, and the multiple first velocities of the target obstacle at the L-1th future time, wherein the multiple boundary information, the multiple first velocities, and the multiple second decelerations correspond one-to-one; fuse the multiple second decelerations of the vehicle at the L-1th future time to obtain the fused second deceleration; determine the safe area of ​​the vehicle at the Lth future time based on the fused second deceleration of the vehicle at the L-1th future time; and determine at least one planned trajectory point corresponding to the Lth future time based on the points within the safe area of ​​the vehicle at the Lth future time.

[0124] In one possible implementation, the trajectory planning module 530 fuses multiple second decelerations corresponding to the vehicle at the (L-1)th future time to obtain the fused second deceleration. Then, it: obtains the weight corresponding to each second deceleration, where the weight is related to the prediction probability corresponding to the predicted trajectory information of the first velocity corresponding to the second deceleration; and performs a weighted calculation on the multiple second decelerations corresponding to the vehicle at the (L-1)th future time based on the weights corresponding to each second deceleration to obtain the fused second deceleration.

[0125] In one possible implementation, the predicted trajectory information also includes multiple first decelerations corresponding to each future time. Before the trajectory planning module 530 fuses the multiple second decelerations corresponding to the vehicle at the (L-1)th future time to obtain the fused second deceleration, it is further configured to: obtain the first deceleration of the target obstacle at the (L-1)th future time; based on the first deceleration of the target obstacle at the (L-1)th future time, correct the second deceleration of the vehicle at the (L-1)th future time to obtain the corrected second deceleration of the vehicle at the (L-1)th future time; when the trajectory planning module 530 fuses the multiple second decelerations corresponding to the vehicle at the (L-1)th future time to obtain the fused second deceleration, it is configured to: fuse the multiple corrected second decelerations corresponding to the vehicle at the (L-1)th future time to obtain the fused second deceleration.

[0126] In one possible implementation, the trajectory planning module 530 corrects the vehicle's second deceleration at the (L-1)-th future time based on the first deceleration of the target obstacle at the (L-1)-th future time, obtaining the corrected second deceleration of the vehicle at the (L-1)-th future time. Then, it determines the target time distance based on the planned trajectory point of the vehicle at the (L-1)-th future time and the predicted trajectory point of the target obstacle at the (L-1)-th future time. The target time distance represents the distance the vehicle travels from the planned trajectory point at the (L-1)-th future time to the predicted trajectory point of the target obstacle at the (L-1)-th future time. The time required to predict the trajectory point; based on the target time distance, the first deceleration of the target obstacle at the L-1 future time is adjusted to obtain the adjusted first deceleration of the target obstacle at the L-1 future time, wherein the adjusted first deceleration is not greater than the original first deceleration, and the target time distance is positively correlated with the adjustment range of the first deceleration; based on the adjusted first deceleration of the target obstacle at the L-1 future time, the second deceleration of the vehicle at the L-1 future time is corrected to obtain the corrected second deceleration of the vehicle at the L-1 future time.

[0127] In one possible implementation, when the trajectory planning module 530 selects one of the multiple planned trajectory information as the target planned trajectory information for the vehicle, it is used to: determine the total cost corresponding to each planned trajectory information among the multiple planned trajectory information, wherein the total cost corresponding to the planned trajectory information is positively correlated with the cost corresponding to each future time, and for any future time, the cost corresponding to the future time is determined based on the distance between the planned trajectory point corresponding to the future time and the boundary indicated by the boundary information corresponding to the future time, which is used to represent the collision risk between the planned trajectory point corresponding to the future time and the boundary indicated by the boundary information corresponding to the future time; and select the planned trajectory information with the minimum corresponding total cost as the target planned trajectory information for the vehicle.

[0128] In one possible implementation, the boundary information includes a first boundary and a second boundary, wherein the distance between the first boundary and the current position of the vehicle is less than the distance between the second boundary and the current position of the vehicle. The trajectory planning module 530 is further configured to: for any future time, determine a first-generation value based on the distance between the planned trajectory point corresponding to the future time and the first boundary corresponding to the future time, and determine a second-generation value based on the distance between the planned trajectory point corresponding to the future time and the second boundary corresponding to the future time; obtain the weights corresponding to the first-generation value and the second-generation value, wherein the weight corresponding to the first-generation value is less than the weight corresponding to the second-generation value; and for any future time, perform a weighted calculation based on the first-generation value, the weight corresponding to the first-generation value, the second-generation value, and the weight corresponding to the second-generation value to determine the generation value corresponding to the future time.

[0129] It should be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the devices and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. In the several embodiments provided in this application, the coupling between modules can be electrical. Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0130] This application also provides an electronic device 60, please refer to... Figure 6 The device includes a processor 610 and a memory 620. The memory 610 stores computer programs, and the processor 620 executes the programs stored in the memory 610 to implement the trajectory planning method described in any embodiment of this application. The electronic device 60 may be, for example, a vehicle, an in-vehicle terminal, or a server.

[0131] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the trajectory planning method described in any embodiment of this application.

[0132] In this application, unless otherwise expressly defined, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0133] The terms “first,” “second,” “third,” “fourth,” etc., in this application (if any) are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. “Multiple” means no fewer than two.

[0134] In this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this application generally indicates that the preceding and following related objects have an "or" relationship. "Multiple" refers to no fewer than two.

[0135] Unless otherwise specified, all steps in this application may be performed sequentially or randomly. For example, if a method includes steps A and B, it means that the method may include steps A and B performed sequentially, or it may include steps B and A performed sequentially. For example, if a method may also include step C, it means that step C may be added to the method in any order. For example, the method may include steps A, B, and C, or it may include steps A, C, and B, or it may include steps C, A, and B, etc.

[0136] The above are merely preferred embodiments of this application and are not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A trajectory planning method, characterized in that, include: Acquire multiple predicted trajectory information corresponding to each target obstacle. Each predicted trajectory information includes multiple predicted trajectory points corresponding to each future time. The target obstacles include obstacles within a preset range of the vehicle. Based on the multiple predicted trajectory information corresponding to the target obstacle, multiple boundary information corresponding to the target obstacle at each of the future times is determined, wherein the multiple boundary information corresponding to the future times is determined based on the multiple predicted trajectory points corresponding to the future times respectively; Based on multiple boundary information corresponding to the target obstacle at each of the future times, multiple planned trajectory information of the vehicle is determined, and each planned trajectory information includes the planned trajectory points corresponding to each of the multiple future times. Select one of the multiple planned trajectory information as the target planned trajectory information of the vehicle.

2. The method according to claim 1, characterized in that, The boundary information includes a first boundary and a second boundary. The distance between the first boundary and the current position of the vehicle is less than the distance between the second boundary and the current position of the vehicle. Determining the multiple boundary information corresponding to the target obstacle at each of the future time moments based on the multiple predicted trajectory information corresponding to the target obstacle includes: Based on the multiple predicted trajectory information corresponding to the target obstacle, a spatiotemporal map is generated. The spatiotemporal map is used to represent multiple occupied areas corresponding to the target obstacle at each of the multiple future times. Each occupied area is determined based on a predicted trajectory point in one of the predicted trajectory information. The occupied area is located in the direction of travel of the vehicle. For any one of the plurality of future moments, the second boundary is determined based on the overlapping portion between the plurality of occupied regions corresponding to the future moment, and the first boundary is determined based on the non-overlapping portion corresponding to the future moment, wherein the non-overlapping portion includes the portion of the total region formed by the plurality of occupied regions other than the overlapping portion.

3. The method according to claim 1, characterized in that, The determination of multiple planned trajectory information of the vehicle based on multiple boundary information corresponding to the target obstacle at each of the future times includes: For the first future moment among the plurality of future moments, based on the current position of the vehicle and the multiple boundary information of the target obstacle corresponding to the first future moment, at least one planned trajectory point corresponding to the first future moment is determined; For the Lth future time among the plurality of future times, based on the planned trajectory point of the vehicle at the (L-1)th future time and the multiple boundary information of the target obstacle at the Lth future time, at least one planned trajectory point corresponding to the Lth future time is determined, where L is an integer not less than 2. Based on the planned trajectory points corresponding to each of the multiple future moments, multiple planned trajectory information of the vehicle is determined.

4. The method according to claim 3, characterized in that, The predicted trajectory information also includes a first speed corresponding to each of multiple future time moments. The step of determining at least one planned trajectory point corresponding to the Lth future time moment based on the planned trajectory point of the vehicle at the (L-1)th future time moment and multiple boundary information of the target obstacle at the Lth future time moment includes: Based on the planned trajectory point of the vehicle at the (L-1)th future time, the second speed of the vehicle at the (L-1)th future time, multiple boundary information of the target obstacle at the Lth future time, and multiple first speeds of the target obstacle at the L-1th future time, multiple second decelerations of the vehicle at the L-1th future time are determined, and the multiple boundary information, the multiple first speeds, and the multiple second decelerations correspond one-to-one; The multiple second decelerations of the vehicle at the (L-1)th future time are fused to obtain the fused second deceleration. Based on the second deceleration of the vehicle after fusion at the (L-1)th future time, the safe area of ​​the vehicle at the Lth future time is determined; Based on the points of the vehicle within the safe area corresponding to the Lth future time, at least one planned trajectory point corresponding to the Lth future time is determined.

5. The method according to claim 4, characterized in that, The step of fusing multiple second decelerations corresponding to the vehicle at the (L-1)th future time to obtain the fused second deceleration includes: Obtain the weight corresponding to each second deceleration. The weight corresponding to the second deceleration is related to the prediction probability corresponding to the prediction trajectory information of the first velocity corresponding to the second deceleration. Based on the weights corresponding to each second deceleration, the multiple second decelerations of the vehicle at the (L-1)th future time are weighted and calculated to obtain the fused second deceleration.

6. The method according to claim 4, characterized in that, The predicted trajectory information also includes multiple first decelerations corresponding to each future time. Before fusing the multiple second decelerations corresponding to the vehicle at the (L-1)th future time to obtain the fused second deceleration, the method further includes: Obtain the first deceleration of the target obstacle at the (L-1)th future time. Based on the first deceleration of the target obstacle at the (L-1)th future time, the second deceleration of the vehicle at the (L-1)th future time is corrected to obtain the corrected second deceleration of the vehicle at the (L-1)th future time. The step of fusing multiple second decelerations corresponding to the vehicle at the (L-1)th future time to obtain the fused second deceleration includes: The multiple corrected second decelerations of the vehicle at the (L-1)th future time are fused to obtain the fused second deceleration.

7. The method according to claim 6, characterized in that, The step of correcting the second deceleration of the vehicle at the (L-1)th future time based on the first deceleration of the target obstacle at the (L-1)th future time to obtain the corrected second deceleration of the vehicle at the (L-1)th future time includes: Based on the planned trajectory point of the vehicle at the (L-1)th future time and the predicted trajectory point of the target obstacle at the (L-1)th future time, the target time distance is determined. The target time distance is used to represent the time required for the vehicle to travel from the planned trajectory point at the (L-1)th future time to the predicted trajectory point of the target obstacle at the (L-1)th future time. Based on the target time distance, the first deceleration of the target obstacle at the (L-1)th future time is adjusted to obtain the adjusted first deceleration of the target obstacle at the (L-1)th future time. The adjusted first deceleration is not greater than the first deceleration before adjustment, and the target time distance is positively correlated with the adjustment range of the first deceleration. Based on the adjusted first deceleration of the target obstacle at the (L-1)th future time, the second deceleration of the vehicle at the (L-1)th future time is corrected to obtain the corrected second deceleration of the vehicle at the (L-1)th future time.

8. The method according to claim 1, characterized in that, Selecting one of the multiple planned trajectory information as the target planned trajectory information of the vehicle includes: The total cost value corresponding to each planning trajectory information in the plurality of planning trajectory information is determined. The total cost value corresponding to the planning trajectory information is positively correlated with the cost value corresponding to each future time. For any future time, the cost value corresponding to the future time is determined based on the distance between the planning trajectory point corresponding to the future time and the boundary indicated by the boundary information corresponding to the future time, and is used to represent the collision risk between the planning trajectory point corresponding to the future time and the boundary indicated by the boundary information corresponding to the future time. The planned trajectory information with the minimum total cost is taken as the target planned trajectory information of the vehicle.

9. The method according to claim 8, characterized in that, The boundary information includes a first boundary and a second boundary, wherein the distance between the first boundary and the current position of the vehicle is less than the distance between the second boundary and the current position of the vehicle, and the method further includes: For any future moment, the first generation value is determined based on the distance between the planned trajectory point corresponding to the future moment and the first boundary corresponding to the future moment, and the second generation value is determined based on the distance between the planned trajectory point corresponding to the future moment and the second boundary corresponding to the future moment. Obtain the weights corresponding to the first-generation value and the second-generation value, where the weights corresponding to the first-generation value are less than the weights corresponding to the second-generation value. For any future moment, the generation value corresponding to the future moment is determined by weighted calculation based on the first generation value corresponding to the future moment, the weight corresponding to the first generation value, the second generation value corresponding to the future moment, and the weight corresponding to the second generation value.

10. A trajectory planning device, characterized in that, include: The acquisition module is used to acquire multiple predicted trajectory information corresponding to each target obstacle. Each predicted trajectory information includes multiple predicted trajectory points corresponding to each future time. The target obstacles include obstacles within a preset range of the vehicle. A boundary determination module is used to determine multiple boundary information corresponding to the target obstacle at each of the future times based on the multiple predicted trajectory information corresponding to the target obstacle. The multiple boundary information corresponding to the future times is determined based on the multiple predicted trajectory points corresponding to the future times. The trajectory planning module is used to determine multiple planned trajectory information of the vehicle based on multiple boundary information corresponding to the target obstacle at each of the future times. Each planned trajectory information includes the planned trajectory points corresponding to each of the multiple future times. Select one of the multiple planned trajectory information as the target planned trajectory information of the vehicle.

11. An electronic device, characterized in that, Includes processor and memory, of which: Memory, used to store computer programs; A processor for executing a program stored in memory to implement the method of any one of claims 1-9.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1-9.

Images