WO2022246750A1

WO2022246750A1 - Trajectory generation method and apparatus, movable platform, and storage medium

Info

Publication number: WO2022246750A1
Application number: PCT/CN2021/096468
Authority: WO
Inventors: 高飞; 纪佳林
Original assignee: 深圳市大疆创新科技有限公司
Priority date: 2021-05-27
Filing date: 2021-05-27
Publication date: 2022-12-01

Abstract

A trajectory generation method and apparatus, a movable platform, and a storage medium. The method comprises: determining a safe passage without an obstacle (S101); checking a trajectory segment in a currently generated trajectory located outside the safe passage (S102); generating a target trajectory point in the safe passage according to the trajectory segment (S103); and using the target trajectory point to optimize the trajectory until the optimized trajectory is within the safe passage (S104). The method can effectively reduce the trajectory optimization dimension and improve the trajectory optimization efficiency.

Description

Trajectory generation method, device, movable platform and storage medium

technical field

The present application relates to the technical field of path planning, and in particular, to a trajectory generation method, device, movable platform, and storage medium.

Background technique

In today's society, mobile platforms are more and more widely used, such as sweeping robots, industrial robotic arms, logistics transportation robots, unmanned aerial vehicles or unmanned vehicles, etc., which have brought a lot of convenience to people's lives. No matter what kind of mobile platform it is, as long as it involves movement, it is necessary to consider the problem of obstacle avoidance and smoothness of movement during the movement.

When the mobile platform performs a task, it must first plan a safe passage. The planned safe passage is enough to avoid all obstacles, and then generate a guide on the basis of this known safe passage to guide the movement of the mobile platform. traces of. Wherein, the safety channel is composed of a plurality of safety areas. In the related art, after obtaining the safety channel, a trajectory expressed by a section of Bezier curve is correspondingly generated in each safety area in the safety channel, and then by solving the QCQP (Quadratically constrained quadratic program, quadratic constrained quadratic programming) to optimize the trajectory expressed by the Bezier curve. In this method, when there are many security areas, the optimization dimension will increase accordingly, which requires more computing resources and has low optimization efficiency.

Contents of the invention

In view of this, one of the objectives of the present application is to provide a trajectory generation method, device, movable platform and storage medium.

In the first aspect, the embodiment of the present application provides a trajectory generation method, including:

Determine a safe passage without obstacles;

Checking the trajectory segments that are outside the safe passage in the currently generated trajectory;

generating a target trajectory point in the safe channel according to the trajectory segment;

Optimizing the trajectory using the target trajectory point until the optimized trajectory is within the safe passage.

In a second aspect, the embodiment of the present application provides a trajectory generation device, including:

memory for storing executable instructions;

one or more processors;

Wherein, when the one or more processors execute the executable instructions, they are individually or jointly configured to:

Determine a safe passage without obstacles;

In a third aspect, the embodiment of the present application provides a mobile platform, including:

body;

a power system, arranged in the fuselage, for driving the movement of the movable platform;

And, the trajectory generation device according to the second aspect.

In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, the computer-readable storage medium stores executable instructions, and when the executable instructions are executed by a processor, the method as described in the first aspect is implemented .

In the trajectory generation method, device, movable platform and storage medium provided in the embodiments of the present application, after a safe passage without obstacles is determined, the currently generated trajectory is checked for the trajectory segment outside the safe passage, Then generate a target trajectory point in the safe passage according to the trajectory segment, and optimize the trajectory by using the target trajectory point until the optimized trajectory is in the safe passage. In this embodiment, it is only necessary to insert target trajectory points in the safety passage using a dynamic point insertion method based on the trajectory segment outside the safety passage, and use the inserted target trajectory points to optimize the trajectory. trajectories, the number of trajectory segments outside the safe passage gradually decreases until all optimized trajectories are within the safe passage. The trajectory generation method provided by this embodiment has nothing to do with the number of safe areas in the safe passage. Optimizing the trajectory corresponding to the safe region in the safe area, but using the inserted target trajectory points to optimize the overall trajectory can effectively reduce the trajectory optimization dimension and improve the trajectory optimization efficiency.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings that need to be used in the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application. For those skilled in the art, other drawings can also be obtained based on these drawings without any creative effort.

Figure 1 is a schematic diagram of a safe channel provided by an embodiment of the present application;

Fig. 2 is a schematic diagram of generating a trajectory in a safe passage provided by an embodiment of the present application;

Fig. 3 is a schematic flowchart of a trajectory generation method provided by an embodiment of the present application;

Fig. 4 is a schematic diagram of a safety channel generated within the historical perception range of a depth sensor provided by an embodiment of the present application;

Fig. 5 is a schematic diagram of a reliable area provided by an embodiment of the present application;

Fig. 6 is a schematic diagram of the growth of a path search tree provided by an embodiment of the present application;

FIG. 7 is a schematic diagram of determining a leaf node safety factor provided by an embodiment of the present application;

Fig. 8 is a schematic diagram of scoring a path search tree provided by an embodiment of the present application;

Fig. 9 is a schematic diagram of a path search tree pruning provided by an embodiment of the present application;

FIG. 10 is a schematic diagram of determining a safe channel based on a path search tree provided by an embodiment of the present application;

Fig. 11 is a schematic diagram of a trajectory optimization provided by an embodiment of the present application;

FIG. 12 is a schematic diagram of the quantitative relationship between target trajectory points and safe areas in a safe passage provided by an embodiment of the present application;

Fig. 13 is a schematic structural diagram of a trajectory generation device provided by an embodiment of the present application;

Fig. 14 is a schematic structural diagram of a mobile platform provided by an embodiment of the present application.

Detailed ways

The following will clearly and completely describe the technical solutions in the embodiments of the application with reference to the drawings in the embodiments of the application. Apparently, the described embodiments are only some of the embodiments of the application, not all of them. Based on the embodiments in this application, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the scope of protection of this application.

When mobile platforms (such as UAVs, mobile robots, etc.) perform mobile tasks in complex environments, they generally need to use depth sensors, such as depth cameras or laser radars, to obtain relative distance information of obstacles in the environment. This information generally gives the projected distance of the actual obstacle in the viewing direction, so it is also called depth information. Depth information is the information necessary to ensure maneuver safety when mobile platforms (such as UAVs, mobile robots, etc.) perform tasks.

The movable platform maintains depth information about obstacles in the current detection environment collected by the depth sensor, so that a safe passage that can avoid obstacles can be subsequently determined based on the depth information maintained by the movable platform. As an example, the depth information includes, but is not limited to, a grayscale image corresponding to a depth value, a color image corresponding to a depth value, a depth image or a point cloud frame.

In a possible implementation, the mobile platform will establish a unified spatial map representing the obstacles in the detection environment based on the above-mentioned depth information about obstacles, such as a three-dimensional grid map, an octree map or a European map. In this method, the depth information of each frame collected by the depth sensor on the movable platform and the existing obstacle information on the map are reasonably calculated and fused to obtain the map obstacle distribution in the sense of maximum likelihood probability. Subsequently, the obstacle query interface of the map can be called to obtain whether the queried spatial point is occupied by obstacles, so as to determine the safe passage that can avoid obstacles. In another possible implementation, the movable platform can project the depth information of each frame collected by the depth sensor into a three-dimensional space to obtain a three-dimensional point cloud about obstacles, and store the three-dimensional point cloud data in a tree-like data structure , such as building a k-d tree or R tree based on the 3D point cloud of each frame, and providing an interface for nearest neighbor query, and then querying the obstacles closest to the spatial point to be measured through the provided interface, and determining the distance between the spatial point to be measured and The distance to the nearest obstacle is used to determine the safe passage that can avoid the obstacle.

For example, please refer to Figure 1. In the environment shown in Figure 1, the mobile platform uses the above-mentioned spatial map or tree data structure to detect the nearest neighbor obstacle, so as to obtain the multiple circular security Area composed of safe passages. In the related art, after obtaining the safety channel, a trajectory expressed by a Bezier curve is correspondingly generated in each circular safety area in the safety channel, and then by solving the QCQP (Quadratically constrained quadratic program, quadratically constrained quadratic program, quadratically constrained quadratic program ) to optimize the trajectory expressed by the Bezier curve. In this method, when there are many security areas, the optimization dimension will increase accordingly, which requires more computing resources and has low optimization efficiency.

Aiming at the problems in the related art, the embodiment of the present application provides a trajectory generation method, after determining an obstacle-free safe passage, check the trajectory segments in the currently generated trajectory that are outside the safe passage, and then according to The trajectory segment generates a target trajectory point in the safety passage, and the trajectory is optimized using the target trajectory point until the optimized trajectory is in the safety passage. In this embodiment, it is only necessary to insert target trajectory points in the safety passage using a dynamic point insertion method based on the trajectory segment outside the safety passage, and use the inserted target trajectory points to optimize the trajectory. trajectories, the number of trajectory segments outside the safe passage gradually decreases until all optimized trajectories are within the safe passage. The trajectory generation method provided by this embodiment has nothing to do with the number of safe areas in the safe passage. Optimizing the trajectory corresponding to the safe region in the safe area, but using the inserted target trajectory points to optimize the overall trajectory can effectively reduce the trajectory optimization dimension and improve the trajectory optimization efficiency.

In some embodiments, the trajectory generation method provided in the embodiment of the present application can be applied to a trajectory generation device.

On the one hand, the trajectory generation device may be a mobile platform with data processing capability. Wherein, examples of the mobile platform include but are not limited to unmanned aerial vehicle, unmanned vehicle, cloud platform, unmanned boat, mobile robot, sweeping robot, industrial mechanical arm or logistics transportation robot, etc.

Exemplarily, the trajectory generation device may be a computer software product integrated in the mobile platform, and the computer software product includes an application program capable of executing the trajectory generation method provided by the embodiment of the present application. Exemplarily, the trajectory generation device may be a movable platform including at least a memory and a processor, and the processor in the movable platform may execute the program stored in the memory indicating the trajectory generation method provided by the embodiment of the present application. Executable instructions.

On the other hand, the trajectory generation device can also be a chip or an integrated circuit with data processing capabilities, and the trajectory generation device includes but is not limited to, for example, a central processing unit (Central Processing Unit, CPU), a digital signal processor (Digital Signal Processor) Processor, DSP), application specific integrated circuit (Application Specific Integrated Circuit, ASIC) or off-the-shelf programmable gate array (Field-Programmable Gate Array, FPGA), etc. The obstacle information acquisition device can be installed in electronic equipment.

Exemplarily, taking the mobile platform as an example of an unmanned aerial vehicle, the unmanned aerial vehicle at least includes a fuselage, a power system installed in the fuselage, and a flight controller; the power system is used to drive the The unmanned aerial vehicle moves, and the flight controller is used to control the flight of the unmanned aerial vehicle; wherein, the trajectory generating device can be a flight controller in the unmanned aerial vehicle, that is, the flight controller in the unmanned aerial vehicle can Execute the trajectory generation method provided by the embodiment of the present application.

Exemplarily, the trajectory generation method provided in this application can be applied in the fields of electric power inspection, fire rescue or security protection. After the trajectory is generated using the method provided in the embodiment of the application, the movable platform can be controlled to move according to the trajectory, so as to realize The intelligent search and inspection process significantly reduces labor costs and improves inspection efficiency.

In an exemplary application scenario, please refer to FIG. 2 , an unmanned aerial vehicle is used as an example for illustration: the unmanned aerial vehicle is equipped with a depth sensor, and the depth sensor includes but is not limited to a lidar or an RGBD camera. In the current detection environment, for example, the terminal waypoint to be reached by this trajectory generation task is determined under the operation of the user, and the current position is used as the starting waypoint. The UAV needs to generate A safe trajectory avoiding obstacles.

In the current detection environment, the unmanned aerial vehicle can use its on-board depth sensor to collect depth information about obstacles, and then maintain the depth through a unified spatial map or tree-like data structure (such as k-d tree) information, then sample space points within the sensing range of the depth sensor, use the above-mentioned space map or tree data structure to perform nearest neighbor obstacle detection on the space point, determine the safe space around the space point, and pass several The sampling of space points and the query of the surrounding safe space determine the obstacle-free flight corridor between the starting waypoint and the ending waypoint (in the field of unmanned aerial vehicles, the safe passage is called the flight corridor).

Next, the trajectory generation method provided in this embodiment can be used. After determining an obstacle-free flight corridor, check the trajectory segments in the currently generated trajectory that are outside the flight corridor. The currently generated trajectory is based on the The starting waypoint and the ending waypoint are determined, or the currently generated trajectory is determined based on the starting waypoint, the ending waypoint and the target waypoint generated in the flight corridor last time; then the none The manned aircraft generates a target waypoint in the flight corridor according to the trajectory segment, uses the target waypoint to optimize the trajectory until the optimized trajectory is in the flight corridor, and finally generates a flight path from the initial waypoint A safe trajectory that avoids obstacles until the waypoint is terminated. Further, the trajectory generated by the embodiment of the present application can be applied in the fields of power inspection, fire rescue or security. For example, it can control the unmanned aerial vehicle to fly according to the generated trajectory. In this way, an intelligent search and inspection process is realized, which significantly reduces labor costs and improves inspection efficiency.

Next, the trajectory generation method provided by the embodiment of the present application is described: please refer to FIG. 3, which is a schematic flow chart of a trajectory generation method provided by the embodiment of the present application. The method can be executed by a trajectory generation device. Methods include:

In step S101, a safe passage without obstacles is determined.

In step S102 , check the currently generated trajectory segments that are outside the safe channel.

In step S103, a target trajectory point is generated in the safe channel according to the trajectory segment.

In step S104, the trajectory is optimized using the target trajectory point until the optimized trajectory is within the safe channel.

For step S101, the safe passage may be a passage determined by the device between the starting track point and the ending track point, which can avoid obstacles. The starting track point and the ending track point can be specifically set according to the actual application scenario, and this embodiment does not impose any restrictions on this; for example, in the power inspection scene, the starting track can be determined according to the distribution of electric wires point and end track point; or in a security scene, the start track point and end track point can be determined according to the distribution of the area to be inspected; or the start track point and end track point can also be user Specify according to the actual situation.

In some embodiments, the safe channel is obtained based on path search between the starting track point and the ending track point. Exemplarily, the device uses the starting track point as a root node to determine a path search tree extending to the end track point, and then determines a target leaf node from all leaf nodes in the path search tree, and finally according to A secure path from the target leaf node to the root node determines the secure channel. It can be understood that the present application does not impose any restrictions on the search algorithm used in the path search process, and can be specifically set according to actual application scenarios. For example, the path search tree can be a forward spanning tree, RRT (rapidly-exploring random tree , fast search random tree) tree, RRT* tree or FMT (fast marching tree, fast marching number) tree, etc.

Here, the acquisition process of the non-root node in the path search tree is described: taking the device installed on a movable platform as an example, and the movable platform is also equipped with a depth sensor. The non-root nodes in the path search tree are spatial points indicating non-obstacles sampled by the device within the current perception range of the depth sensor.

When sampling the spatial point, in order to improve the validity of the sampled spatial point, the device can obtain the sampling result within the historical perception range of the depth sensor; then estimate the current perception range of the depth sensor according to the sampling result The reliable area of , and then sample the spatial point in the reliable area. In this embodiment, prior information (sampling results within the historical perception range) is used to guide the sampling process within the current perception range, which is conducive to improving the validity of the sampled spatial points and further improving the trajectory generation efficiency.

Of course, in the case where the sampling results within the historical sensing range of the depth sensor cannot be obtained, for example, the current sensing range of the depth sensor is the first sensing process of the depth sensor in this environment, the device can The spatial points are uniformly sampled within the current sensing range of the sensor.

In a possible implementation manner, the reliable area within the current sensing range of the depth sensor is determined according to the safe channel generated by the sampling result. Specifically, the movable platform needs to effectively detect the environment between the starting track point and the ending track point, and obtain obstacle information through a depth sensor. For example, the obstacle information includes The depth information of the depth sensor is used to collect the depth information of the obstacle within the sensing range of the depth sensor in the environment, and the depth information represents the relative distance between the obstacle within the current sensing range and the movable platform. During this detection process, the movable platform gradually moves from the starting track point to the ending track point. With the movement of the movable platform, the environmental content within the sensing range of the depth sensor also gradually changes, thereby achieving the acquisition of the starting track point. Obstacle information of the environment between the start track point and the end track point.

Within each sensing range of the depth sensor, the device performs the following steps: the device samples a plurality of spatial points indicating non-obstacles within the current sensing range of the depth sensor according to the obstacle information collected by the depth sensor ; Then generate a path search tree extending to the end trajectory point according to the starting trajectory point and the plurality of space points, determine the safe passage according to the path search tree, and then use the trajectory generation method provided by the embodiment of the present application Generate a trajectory within the safe passage.

Further, considering that the safety probability of the environment space where the safe passage is located is relatively high, when the device samples space points within the current perception range of the depth sensor, the device can To determine a reliable area within the current sensing range, wherein the historical sensing range overlaps with the current sensing range, so as to ensure that part or all of the environmental space where the safe passage corresponding to the historical sensing range is located is also within the current sensing range of the depth sensor, so that the device can effectively determine the reliable area based on the safe passage corresponding to the historical sensing range. Exemplarily, the reliable area includes at least the safe passage corresponding to the historical sensing range environment space. Exemplarily, the device may effectively determine the reliable area according to the safe channel corresponding to one or more historical sensing ranges, and the one or more historical sensing ranges and the current sensing range all have overlapping areas.

In an example, please refer to Figure 4, assuming that at time t, the safe channel generated by the sampling results within the historical perception range FoV(t) of the depth sensor is shown in Figure 4, at time t+1, the depth sensor’s Within the current perception range FoV(t+1), the device can determine the reliable area as shown in Figure 5 according to the environmental space where the safe passage is located as shown in Figure 4, and the safety probability of the reliable area is compared with other areas The security probability of is higher, where the historical perception range FoV(t) and the current perception range FoV(t+1) have an overlapping area.

Furthermore, after determining the reliable area within the current range, since the safety probability of the reliable area is higher than that of other areas, the device may sample spatial points at a high sampling rate in the reliable area, And, sampling spatial points at a low sampling rate in the unreliable area of the current perception range. In this embodiment, the spatial point sampling process can be guided according to prior information, effectively improving the validity of the sampled spatial points, improving sampling efficiency and trajectory generation efficiency, and because the reliable area is based on the historical perception range The safety channel generated by the sampling results is determined, which indirectly improves the overlapping range of the trajectory generated in the current perception range and the trajectory generated in the historical perception range, thereby improving the consistency with the perception information/historical planning trajectory. After sampling the space point within the current sensing range of the depth sensor, the device also needs to check the safety of the space point. As mentioned above, the depth sensor is used to collect the depth information of obstacles within the sensing range of the depth sensor in the environment, and the depth information represents the relative distance between the obstacle within the current sensing range and the movable platform, and then the The device will maintain the depth information of obstacles collected by the depth sensor in the current sensing range and/or the historical sensing range; as an example, the current sensing range and/or the The depth information of obstacles collected in the historical perception range constructs a unified spatial map such as a grid map; as an example, a tree-like data structure can be used to store the depth information of obstacles collected in each perception range. The depth information of obstacles collected within the perception range is projected into a three-dimensional space to obtain a three-dimensional point cloud, and a k-d tree or R tree is established using the three-dimensional point cloud.

Therefore, after the spatial point is sampled within the current sensing range of the depth sensor, the device can determine the distance between the spatial point and the nearest The distance between adjacent obstacles realizes the safety inspection of the space point; wherein, the obstacle information includes the depth information about obstacles collected by the depth sensor within its sensing range, exemplary , the obstacle information also includes k-d tree, R tree and/or grid map constructed according to the depth information; and then the device can reserve the distance between the spatial point and the nearest neighbor obstacle greater than The spatial point of the preset distance threshold, wherein the preset distance threshold can be specifically set according to the actual application scenario, and this embodiment does not make any restrictions on this, for example, the preset distance threshold can be set according to the movable platform Size OK.

After the spatial point is sampled within the current perception range of the depth sensor and the safety of the spatial point is determined, the spatial point may be added to the path search tree as a node of the path search tree. Exemplarily, in order to speed up the generation of the path search tree, the spatial points can be added to the path search tree according to the distance from the root node from near to far, so as to quickly obtain the Path search tree. Exemplarily, the spatial points may also be randomly added to the path search tree, so as to obtain a path search tree extending toward the termination track point.

In some embodiments, it is considered that the generated path search tree may appear as shown in FIG. 6 , that is, the leaf nodes of the path search tree grow toward obstacles, resulting in short-sighted appearance of local planning and falling into a dead end. Therefore, in order to improve the safety of the subsequently determined safety channel, after determining the path search tree extending to the termination track point, the device needs to perform pruning processing on the path search tree, and the scene shown in Figure 6 will appear The leaf nodes or branches are cut off. Exemplarily, the device determines the safety factor of each leaf node according to the relationship between each leaf node and the nearest neighbor obstacle in the path search tree; and then traverses the path search tree starting from each leaf node, Determine the safety factor of each non-leaf node; finally, perform pruning processing on the path search tree according to the safety factor of each node in the path search tree. In this embodiment, by performing security assessment on the nodes in the path search tree, pruning is performed according to the result of the security assessment (i.e., the safety factor), so that most useless nodes can be quickly cut off, leaving a real feasible area for expressing the environment branches.

Wherein, the nearest neighbor obstacle of each leaf node is obtained from pre-stored obstacle information based on the location of the leaf node; the obstacle information is obtained by the device in the current sensing range of the depth sensor and/or Obtained within the historical sensing range; the obstacle information includes depth information about obstacles collected by the depth sensor within its sensing range. Exemplarily, the obstacle information also includes Constructed k-d trees, R trees and/or raster maps.

After determining the nearest neighbor obstacle of each leaf node, the device can determine the safety factor of each leaf node in the following manner:

In a possible implementation manner, for each leaf node, the device determines the direction of the leaf node towards the nearest neighbor obstacle, and then according to the angle between the direction and the growth direction of the leaf node , to determine the safety factor of the leaf node; wherein, the direction of the leaf node towards the nearest neighbor obstacle may be the direction of the perpendicular (or the shortest line, etc.) between the leaf node and the obstacle; The growth direction of the leaf node is the direction of the connecting line between the leaf node and its parent node.

Wherein, the safety factor of the leaf node is positively correlated with the included angle, please refer to Fig. 7, the larger the included angle α, it means that if the leaf node continues to grow, the growth direction of the branch will be different from the distance of the nearest neighbor obstacle. If the collision probability is low, the security of the leaf node is higher, and the corresponding safety factor is larger; on the contrary, the smaller the angle α is, it means that if the leaf node continues to grow, the growth direction of the branch is the same as that of the nearest neighbor. The higher the collision probability of obstacles, the lower the safety of the leaf node, and the smaller the corresponding safety factor.

Exemplarily, after obtaining the included angle of the leaf node, if the included angle is greater than a preset angle threshold, determine that the safety factor of the leaf node is a first preset value; otherwise, at the included angle If it is not greater than the preset angle threshold, it is determined that the safety factor of the leaf node is a second preset value, and the first preset value is greater than the second preset value. It can be understood that the preset angle threshold, the first preset value and the second preset value can be specifically set according to the actual application scenario, and this embodiment does not make any limitation thereto; as an example, the preset angle The threshold value is greater than 90°, the first preset value is 1, and the second preset value is 0.

In a second possible implementation manner, for each leaf node, the device may determine the safety factor of the leaf node according to the distance between the leaf node and the nearest neighbor obstacle. Wherein, the safety factor of the leaf node is positively correlated with the distance, that is, the farther the distance is, the higher the safety of the leaf node is, and the closer the distance is, the lower the safety factor of the leaf node is.

In some embodiments, after determining the safety factor of each leaf node, the device may traverse the path search tree starting from each leaf node, and determine the safety factor of each non-leaf node. Exemplarily, the safety factor of the non-leaf node is the sum of the safety factors of its child nodes. For example, please refer to FIG. 8 , the safety factor of each leaf node is 0 or 1, and the safety factor of each non-leaf node is the sum of the safety factors of its child nodes.

After determining the safety factor of each node in the path search tree, the device may perform pruning processing on the path search tree according to the safety factor of each node in the path search tree; Nodes in the path search tree whose safety factor is lower than a preset threshold are pruned. The preset threshold can be specifically set according to actual application scenarios. In this embodiment, considering that if the nodes whose safety factor is lower than the preset threshold continue to be generated, the growth direction of the branch will have a higher collision probability with the nearest neighbor obstacle, so the nodes whose safety factor is lower than the preset threshold are cut off, leaving The branches that really express the feasible area of the environment are lowered, which is conducive to ensuring the accuracy of the subsequent safe passage.

In some embodiments, in order to avoid the influence of error factors, after determining the safety factor of each node in the path search tree, the safety factor of each node in the path search tree can be normalized, and the The safety factor of each node in the path search tree is mapped to a preset range; exemplary, the normalization process includes: for each node, comparing its own safety factor with the maximum safety factor among nodes belonging to the same parent node The ratio between was determined as the normalized value. After the normalization processing, the device may perform pruning processing on nodes in the path search tree whose safety factor is lower than a preset threshold. The preset threshold can be specifically set according to actual application scenarios.

As an example, please refer to Figure 9. Figure 9 is the result of normalizing the path search tree shown in Figure 8. For example, for a node with a safety factor of 0.25, its normalized safety factor of 0.25 is itself The ratio of the original safety factor of 1 to a safety factor of 4, the node with a safety factor of 4 belongs to the same parent node with the node and has the largest safety factor; and after the normalization process, the device can search the tree for the path Nodes whose middle safety factor is lower than the preset threshold (for example, lower than 0.3) are pruned, and the path search tree shown in Figure 9 is obtained, where the dotted line part is the useless node or branch cut off this time, thus leaving Branches that really express the feasible region of the environment.

After determining the path search tree, the device determines a target leaf node from all leaf nodes in the path search tree; and then determines the safe channel according to a safe path from the target leaf node to the root node. It can be understood that this embodiment does not impose any restrictions on the selection of the target leaf node, and specific settings can be made according to the actual application scenario; for example, in order to extend to the end track point more quickly, all leaf nodes can be Select the leaf node with the shortest distance from the termination track point as the target leaf node, that is, the distance between the target leaf node and the termination track point is the shortest; of course, the target node can also be selected based on other strategies, For example, the distance from the target node to the nearest neighbor obstacle is the farthest.

Exemplarily, after determining the security path from the target leaf node to the root node, the device may generate multiple security areas on the security path, and then acquire a security channel composed of the multiple security areas. As an example, please refer to Figure 10, the center points of the multiple security areas are all in the security path, the center point of one security area is the target leaf node, and the center point of other security areas is the previous security area The intersection with the safety path, and the size of each safety area is determined according to the distance between a part of the safety path corresponding to the safety area and the nearest neighbor obstacle, so as to obtain the multi-safety path as shown in Figure 10. Area composed of safe passages. Wherein, the safety area may be a circle, or other shapes, such as a rectangle, which is not limited in this embodiment.

In steps S102 to S104, after determining a safe passage without obstacles, the device checks the currently generated trajectory for the trajectory segment outside the safe passage; wherein, in the first optimization process, the current The generated trajectory is determined based on the initial trajectory point and the termination trajectory point; in the non-first optimization process, the currently generated trajectory is based on the initial trajectory point, the termination trajectory point and the last time in the safe channel The target trajectory point generated within is determined; then the device can generate a target trajectory point in the safe passage according to the trajectory segment, and use the target trajectory point to optimize the trajectory until the optimized trajectory is in the safe passage Inside. The trajectory generation method provided by this embodiment has nothing to do with the number of safety areas in the safety channel. It is not necessary to optimize the trajectory corresponding to the safety area in each safety area, but to use the inserted target trajectory points to optimize the overall trajectory, which can effectively reduce Trajectory optimization dimension to improve trajectory optimization efficiency.

Exemplarily, when determining the trajectory segment, the device determines the spatial position of the safe passage and the spatial position of the currently generated trajectory, and determines the position of the currently generated trajectory according to the non-overlapping part of the two spatial positions. Trajectory segments that are outside the safe passage.

Here is an exemplary description of the generation of the target track point:

In a first possible implementation manner, the device may determine the target area in the safety passage according to the trajectory segment, for example, two intersection points of the trajectory segment and the safety passage may be respectively determined according to boundaries, and then determine the area within the two boundaries in the safe channel as the target area, and then generate target track points in the target area; The generation of target trajectory points in the relevant target area is beneficial to improve the speed of trajectory optimization; and it is not necessary to generate target trajectory points in the entire safety channel, which is also conducive to improving the efficiency of target trajectory point generation.

Exemplarily, in order to improve the speed of trajectory optimization, the target trajectory point may be generated at the farthest distance from the trajectory segment in the target area, more specifically, the edge of the target area close to the trajectory segment In the position, the target trajectory point is generated at the farthest distance from the trajectory segment, so that the optimized trajectory based on the target trajectory point can be guaranteed to be within the safe channel as much as possible; The target track point is generated at a random position in the area.

In a second possible implementation, the device may generate a plurality of candidate trajectory points in the safe channel, and then, according to the distance between the candidate trajectory points and the trajectory segment, select Select the target track point; Exemplarily, the distance between the target track point and the track segment is the farthest; Exemplarily, the candidate track point with the second farthest distance from the track segment may also be selected as the target track point. In this embodiment, the target trajectory point is selected based on the distance between the candidate trajectory point and the trajectory segment, which is beneficial to improve the efficiency of trajectory optimization.

In an exemplary implementation, the safety channel is composed of a plurality of safety areas, and the device may generate one or more candidate waypoints in the intersection area of the plurality of safety areas. Exemplarily, the candidate trajectory point is located at the edge position of the intersection area close to the side of the trajectory segment, so that on the premise of ensuring that the optimized trajectory is within the safe channel, the optimized trajectory is the same as the pre-optimized trajectory. The difference in trajectory is smaller.

In another exemplary embodiment, the safety channel is composed of multiple safety areas, and the device can determine the target safety area from the multiple safety areas according to the intersection position of the trajectory segment and the safety channel, For example, according to the positions of the two intersection points of the trajectory segment and the safety channel, two intersecting target safety areas can be determined respectively, and then the safety area between the two target safety areas can also be determined as the target safety area ; Then, the device generates a plurality of candidate track points within the target safety area; Exemplarily, the candidate track points are located at an edge position close to the side of the track segment in the intersection area of the target safety area. In this embodiment, choosing to generate candidate trajectory points in the target safety area related to the trajectory segment is conducive to improving the trajectory optimization speed; and it is not necessary to generate candidate trajectory points in the entire safe passage, which is also conducive to improving the generation of target trajectory points. efficiency.

As an example, please refer to Fig. 11. After determining an obstacle-free safe passage, the device uses a dynamic interpolation method to perform trajectory optimization. When optimizing for the first time, the trajectory generated by the starting trajectory point and the ending trajectory point is obtained. , that is, the dotted line part in the safety passage in the upper part of Figure 11, and then check the trajectory segment outside the safety passage in the trajectory, according to the intersection position of the trajectory segment and the safety passage, from multiple Determining a target safe area in a safe area, and then generating a plurality of candidate track points in the target safe area, the candidate track points are located at the edge positions near the side of the track segment in the intersecting area of the target safe area, According to the distance between the candidate track point and the track segment, select the candidate track point with the farthest distance from the plurality of candidate track points as the target track point, and then the device uses the target track point to optimize the According to the above trajectory, the optimized trajectory of the solid line part in the upper part of the safety passage as shown in Figure 11 is obtained.

When optimizing for the second time, check the trajectory segment outside the safety passage in the optimized trajectory (the dotted line part in the safety passage of the lower part in Fig. 11), the trajectory after the optimization is defined by the initial trajectory point , the terminating trajectory point and the trajectory generated by the target trajectory point generated in the safety passage last time; then the device determines the target safety area from a plurality of safety regions according to the intersection position of the trajectory segment and the safety passage, and then A plurality of candidate trajectory points are generated in the target safety area, and the candidate trajectory points are located at an edge position close to the side of the trajectory segment in the intersection area of the target safety area, according to the candidate trajectory points and the trajectory segment distance, select the candidate track point with the farthest distance as the target track point from the plurality of candidate track points, then the device uses the target track point to optimize the track, and obtain the following as shown in Figure 11 Half of the optimized trajectories of the solid-line part of the safety passage, check that all of the optimized trajectories are in the safety passage, and end.

In some embodiments, after experimental data verification, please refer to FIG. 12, which shows the change of the number of target trajectory points that need to be generated as the number of safe areas in the safe passage increases. In a complex environment, the safe passage The number of safe areas in the system is generally about 10. In the related art, it is necessary to optimize a section of trajectory in 10 safe areas, and the trajectory generation method provided by the embodiment of the present application is aimed at the same number of safe areas (such as 10 safe areas) only need to insert 1 or 2 target trajectory points to cover most of the scenarios, which effectively improves the efficiency of trajectory generation and optimization.

Correspondingly, referring to FIG. 13 , the embodiment of the present application also provides a trajectory generation device 200, including:

memory 202 for storing executable instructions;

one or more processors 201;

Wherein, when the one or more processors 201 execute the executable instructions, they are individually or jointly configured to:

Determine a safe passage without obstacles;

In an embodiment, the processor 201 is further configured to: determine an obstacle-free safe passage between the starting track point and the ending track point.

In an embodiment, the currently generated track is determined based on the start track point and the end track point; or, the currently generated track is determined based on the start track point, the end track point and the above A target trajectory point determination generated within the safety channel.

In an embodiment, the processor 201 is further configured to: generate a plurality of candidate trajectory points in the safe channel; Select the target track point in .

In an embodiment, the distance between the target track point and the track segment is the farthest.

In an embodiment, the safe channel is composed of multiple safe areas.

The processor 201 is further configured to: generate a plurality of candidate waypoints in the intersecting areas of the plurality of safety areas.

In an embodiment, the candidate track point is located at an edge position on a side of the intersection area close to the track segment.

In an embodiment, the safe channel is composed of multiple safe areas.

The processor 201 is further configured to: determine a target safe area from multiple safe areas according to the intersection position of the track segment and the safe passage; generate multiple candidate track points in the target safe area.

In an embodiment, the candidate track point is located at an edge position close to the side of the track segment in the intersecting area of the target safety area.

In an embodiment, the processor 201 is further configured to: determine a target area in the safe passage according to the track segment, and generate target track points in the target area; wherein, the boundary of the target area It is determined according to the intersection position of the trajectory segment and the safety channel.

In an embodiment, the processor 201 is further configured to: generate a target track point at a place farthest from the track segment in the target area.

In an embodiment, the safety channel is obtained based on path search between the starting track point and the ending track point.

In an embodiment, the processor 201 is further configured to: use the starting track point as a root node to determine a path search tree extending to the end track point; from all leaf nodes of the path search tree Determine the target leaf node; determine the security channel according to the security path from the target leaf node to the root node.

In an embodiment, the distance between the target leaf node and the termination track point is the shortest.

In an embodiment, the non-root nodes in the path search tree are spatial points indicating non-obstacles sampled within the current sensing range of the depth sensor.

In an embodiment, when determining the spatial point, the processor 201 is further configured to: obtain a sampling result within the historical sensing range of the depth sensor; estimate the current sensing range according to the sampling result A reliable region within ; the spatial point is sampled within the reliable region.

In an embodiment, the processor 201 is further configured to: sample spatial points at a high sampling rate in the reliable region, and sample spatial points at a low sampling rate in an unreliable region of the current perception range.

In an embodiment, the reliable region is determined according to a safe channel generated from the sampling result.

In an embodiment, the historical sensing range and the current sensing range have an overlapping area.

In an embodiment, after the spatial point is sampled in the reliable area, the processor 201 is further configured to: according to the obstacle information obtained in the current sensing range and/or the historical sensing range, Determine the distance between the spatial point and the nearest neighbor obstacle; retain the spatial point whose distance is greater than a preset distance threshold.

In an embodiment, the spatial points are added to the path search tree in descending order of distance from the root node.

In an embodiment, after the determination of the path search tree extending to the end track point, the processor 201 is further configured to: search the path according to the distance between each leaf node and the nearest neighbor obstacle in the path search tree. Relationship, determine the safety factor of each leaf node; traverse the path search tree from each leaf node, determine the safety factor of each non-leaf node; according to the safety factor of each node in the path search tree, the path Search the tree for pruning.

In an embodiment, the safety factor of the non-leaf node is the sum of the safety factors of its child nodes.

In an embodiment, the processor 201 is further configured to: for each leaf node, determine the direction of the leaf node towards the nearest obstacle; according to the distance between the direction and the growth direction of the leaf node The included angle determines the safety factor of the leaf node.

In an embodiment, the safety factor of the leaf node is positively correlated with the included angle.

In one embodiment, if the included angle is greater than a preset angle threshold, the safety factor of the leaf node is a first preset value; otherwise, the safety factor of the leaf node is a second preset value, and the first A preset value is greater than the second preset value.

In one embodiment, the growth direction of the leaf node is the direction of the connection line between the leaf node and its parent node.

In an embodiment, the processor 201 is further configured to: for each leaf node, determine the safety factor of the leaf node according to the distance between the leaf node and the nearest neighbor obstacle.

In an embodiment, the safety factor of the leaf node is positively correlated with the distance.

In an embodiment, before performing pruning on the path search tree according to the safety factor of each node in the path search tree, the processor 201 is further configured to: The safety factor of each node is normalized.

In an embodiment, the normalization process includes: for each node, determining a ratio between its own safety factor and the maximum safety factor among nodes belonging to the same parent node as a normalized value.

In an embodiment, the pruning process includes: performing pruning processing on nodes in the path search tree whose safety factor is lower than a preset threshold.

In an embodiment, the nearest neighbor obstacle is obtained from pre-stored obstacle information based on the location of the leaf node; the obstacle information is within the current sensing range and/or the depth sensor acquired within the context of the historical perception described above.

In an embodiment, the secure channel is composed of multiple secure areas generated in the secure path.

In one embodiment, the center points of the multiple security areas are all in the security path; the center point of one security area is the target leaf node, and the center points of other security areas are the previous security area and the target leaf node. The intersection point of the security path.

In an embodiment, the size of each safety area is determined according to the distance between a part of the safety path corresponding to the safety area and the nearest obstacle.

As for the device embodiment, since it basically corresponds to the method embodiment, for related parts, please refer to the part description of the method embodiment. The device embodiments described above are only illustrative, and the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in One place, or it can be distributed to multiple network elements. Part or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. It can be understood and implemented by those skilled in the art without creative effort.

Various implementations described herein can be implemented using a computer readable medium such as computer software, hardware, or any combination thereof. For hardware implementation, the embodiments described herein can be implemented by using Application Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Digital Signal Processing Devices (DSPDs), Programmable Logic Devices (PLDs), Field Programmable Gate Arrays ( FPGA), processors, controllers, microcontrollers, microprocessors, electronic units designed to perform the functions described herein. For software implementation, an embodiment such as a procedure or a function may be implemented with a separate software module that allows at least one function or operation to be performed. The software codes can be implemented by a software application (or program) written in any suitable programming language, stored in memory and executed by a controller.

In an exemplary embodiment, referring to FIG. 14 , a movable platform 10 is also provided, which includes:

fuselage 100;

The power system 110 is arranged in the fuselage and is used to drive the movement of the movable platform;

And, the trajectory generation device 200 described above.

In one embodiment, the movable platform includes a depth sensor, and the depth sensor is used to collect depth information of obstacles within its sensing range.

In an exemplary embodiment, there is also provided a non-transitory computer-readable storage medium including instructions, such as a memory including instructions, which are executable by a processor of an apparatus to perform the above method. For example, the non-transitory computer readable storage medium may be ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, optical data storage device, and the like.

A non-transitory computer-readable storage medium, enabling the terminal to execute the above method when instructions in the storage medium are executed by a processor of the terminal.

It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is a relationship between these entities or operations. There is no such actual relationship or order between them. The term "comprises", "comprises" or any other variation thereof is intended to cover a non-exclusive inclusion such that a process, method, article or apparatus comprising a set of elements includes not only those elements but also other elements not expressly listed elements, or also elements inherent in such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article or apparatus comprising said element.

The methods and devices provided by the embodiments of the present application have been described in detail above. The principles and implementation methods of the present application have been explained by using specific examples in this paper. The descriptions of the above embodiments are only used to help understand the methods and methods of the present application. core idea; at the same time, for those of ordinary skill in the art, according to the idea of this application, there will be changes in the specific implementation and application scope. In summary, the content of this specification should not be construed as limiting the application .

Claims

A trajectory generation method, characterized in that, comprising:

Determine a safe passage without obstacles;

Checking the trajectory segments that are outside the safe passage in the currently generated trajectory;

generating a target trajectory point in the safe channel according to the trajectory segment;

Optimizing the trajectory using the target trajectory point until the optimized trajectory is within the safe passage.
The method according to claim 1, wherein said determining an obstacle-free safe passage comprises:

Determine an obstacle-free safe passage between the starting track point and the ending track point.
The method according to claim 2, wherein the currently generated track is determined based on the starting track point and the ending track point;

Alternatively, the currently generated trajectory is determined based on the starting trajectory point, the ending trajectory point, and a target trajectory point generated in the safety channel last time.
The method according to claim 1, wherein said generating a target trajectory point in said safety channel according to said trajectory segment comprises:

generating a plurality of candidate trajectory points in the safe channel;

The target track point is selected from the plurality of candidate track points according to the distance between the candidate track point and the track segment.
The method according to claim 4, characterized in that the distance between the target trajectory point and the trajectory segment is the farthest.
The method according to claim 4, wherein the safe channel is composed of a plurality of safe areas;

The determining and generating a plurality of candidate trajectory points in the safe channel includes:

A plurality of candidate waypoints are generated within intersection areas of the plurality of safety areas.
The method according to claim 6, wherein the candidate track point is located at an edge position on a side of the intersection area close to the track segment.
The method according to claim 4, wherein the safe channel is composed of a plurality of safe areas;

The determining and generating a plurality of candidate trajectory points in the safe channel includes:

determining a target safety area from a plurality of safety areas according to the intersection position of the trajectory segment and the safety channel;

A plurality of candidate trajectory points are generated within the target safe area.
The method according to claim 8, wherein the candidate track point is located at an edge position near the side of the track segment in the intersecting area of the target safety area.
The method according to claim 1, wherein said generating a target trajectory point in said safety channel according to said trajectory segment comprises:

Determining a target area in the safety passage according to the trajectory segment, and generating target trajectory points within the target area; wherein, the boundary of the target area is determined according to the intersection position of the trajectory segment and the safety passage.
The method according to claim 10, wherein said generating target track points in said target area comprises:

A target trajectory point is generated at the furthest distance from the trajectory segment in the target region.
The method according to claim 2, wherein the safety channel is obtained based on a path search between the starting track point and the ending track point.
The method according to claim 2 or 12, wherein said determining an obstacle-free safe passage between the starting track point and the ending track point comprises:

Using the starting track point as a root node, determine a path search tree extending to the ending track point;

determining a target leaf node from all leaf nodes of the path search tree;

Determine the safe channel according to the safe path from the target leaf node to the root node.
The method according to claim 13, wherein the distance between the target leaf node and the termination track point is the shortest.
The method according to claim 13, wherein the non-root nodes in the path search tree are spatial points indicating non-obstacles sampled within the current perception range of the depth sensor.
The method according to claim 15, wherein the spatial point is determined in the following manner:

Acquiring sampling results within the historical perception range of the depth sensor;

Estimating a reliable area within the current perception range according to the sampling result;

The spatial points are sampled within the reliable region.
The method according to claim 16, wherein said sampling said spatial points in said reliable region comprises:

The spatial points are sampled at a high sampling rate in the reliable region, and the spatial points are sampled at a low sampling rate in the unreliable region of the current perception range.
The method according to claim 16, wherein the reliable region is determined according to a safe channel generated by the sampling result.
The method according to claim 16, wherein the historical sensing range and the current sensing range have an overlapping area.
The method according to claim 16, wherein after sampling the spatial points in the reliable region, further comprising:

determining the distance between the spatial point and the nearest neighbor obstacle according to the obstacle information obtained within the current sensing range and/or the historical sensing range;

Spatial points whose distance is greater than a preset distance threshold are reserved.
The method according to claim 15, wherein the spatial points are added to the path search tree according to the distance from the root node from near to far.
The method according to claim 13, further comprising: after said determining the path search tree extending to said termination track point:

According to the relationship between each leaf node and the nearest neighbor obstacle in the path search tree, determine the safety factor of each leaf node;

Traversing the path search tree from each leaf node to determine the safety factor of each non-leaf node;

The path search tree is pruned according to the safety factor of each node in the path search tree.
The method according to claim 22, wherein the safety factor of the non-leaf node is the sum of the safety factors of its child nodes.
The method according to claim 22, wherein, according to the relationship between each leaf node in the path search tree and the nearest neighbor obstacle, determining the safety factor of each leaf node includes:

For each leaf node, determine the direction of the leaf node towards the nearest neighbor obstacle;

The safety factor of the leaf node is determined according to the angle between the direction and the growth direction of the leaf node.
The method according to claim 24, characterized in that the safety factor of the leaf node is positively correlated with the included angle.
The method according to claim 24, wherein, if the included angle is greater than a preset angle threshold, the safety factor of the leaf node is a first preset value; otherwise, the safety factor of the leaf node is a second preset value. A preset value, the first preset value is greater than the second preset value.
The method according to claim 24, wherein the growth direction of the leaf node is the direction of the connection line between the leaf node and its parent node.
The method according to claim 22, wherein, according to the relationship between each leaf node in the path search tree and the nearest neighbor obstacle, determining the safety factor of each leaf node includes:

For each leaf node, the safety factor of the leaf node is determined according to the distance between the leaf node and the nearest neighbor obstacle.
The method according to claim 28, wherein the safety factor of the leaf node is positively correlated with the distance.
The method according to claim 22, characterized in that before performing pruning processing on the path search tree according to the safety factor of each node in the path search tree, further comprising:

The safety factor of each node in the path search tree is normalized.
The method according to claim 30, wherein the normalization process comprises: for each node, determining the ratio between its own safety factor and the maximum safety factor among nodes belonging to the same parent node as the normalized Normalized value.
The method according to any one of claims 22 to 31, wherein the pruning process includes: pruning nodes in the path search tree whose safety factor is lower than a preset threshold.
The method according to claim 22, wherein the nearest neighbor obstacle is obtained from pre-stored obstacle information based on the location of the leaf node; the obstacle information is obtained by a depth sensor within its current perception range and/or within historical perception.
The method according to claim 13, wherein the secure channel is composed of a plurality of secure areas generated in the secure path.
The method according to claim 34, wherein the central points of the plurality of safe areas are all in the safe path;

The center point of one of the security areas is the target leaf node, and the center points of the other security areas are the intersection points of the previous security area and the security path.
The method according to claim 34 or 35, characterized in that the size of each safety area is determined according to the distance between a part of the safety path corresponding to the safety area and the nearest obstacle.
A trajectory generating device is characterized in that it comprises:

memory for storing executable instructions;

one or more processors;

Wherein, when the one or more processors execute the executable instructions, they are individually or jointly configured to:

Determine a safe passage without obstacles;

Checking the trajectory segments that are outside the safe passage in the currently generated trajectory;

generating a target trajectory point in the safe channel according to the trajectory segment;

Optimizing the trajectory using the target trajectory point until the optimized trajectory is within the safe passage.
The device according to claim 37, wherein the processor is further configured to: determine an obstacle-free safe passage between the starting track point and the ending track point.
The device according to claim 38, wherein the currently generated track is determined based on the starting track point and the ending track point;

Alternatively, the currently generated trajectory is determined based on the starting trajectory point, the ending trajectory point, and a target trajectory point generated in the safety channel last time.
The device according to claim 37, wherein the processor is further configured to:

generating a plurality of candidate trajectory points in the safe channel;

The target track point is selected from the plurality of candidate track points according to the distance between the candidate track point and the track segment.
The apparatus of claim 40, wherein the distance between the target trajectory point and the trajectory segment is the furthest.
The device according to claim 40, wherein the safe channel is composed of a plurality of safe areas;

The processor is further configured to: generate a plurality of candidate waypoints within the intersecting areas of the plurality of safety areas.
The apparatus according to claim 42, wherein the candidate track point is located at an edge position on a side of the intersection area close to the track segment.
The device according to claim 40, wherein the safe channel is composed of a plurality of safe areas;

The processor is also used to:

determining a target safety area from a plurality of safety areas according to the intersection position of the trajectory segment and the safety channel;

A plurality of candidate trajectory points are generated within the target safe area.
The device according to claim 44, wherein the candidate trajectory point is located at an edge position close to the side of the trajectory segment in the intersecting area of the target safety area.
The device according to claim 37, wherein the processor is further configured to: determine a target area in the safety passage according to the track segment, and generate a target track point in the target area; wherein, The boundary of the target area is determined according to the position of the intersection of the track segment and the safety channel.
The apparatus according to claim 46, wherein the processor is further configured to: generate a target track point at the farthest distance from the track segment in the target area.
The device according to claim 38, wherein the safe channel is obtained based on a path search between the starting track point and the ending track point.
The device according to claim 38 or 48, wherein the processor is further configured to:

Using the starting track point as a root node, determine a path search tree extending to the ending track point;

determining a target leaf node from all leaf nodes of the path search tree;

Determine the safe channel according to the safe path from the target leaf node to the root node.
The device according to claim 49, wherein the distance between the target leaf node and the termination track point is the shortest.
The device according to claim 49, wherein the non-root nodes in the path search tree are spatial points indicating non-obstacles sampled within the current sensing range of the depth sensor.
The device according to claim 51, wherein when determining the spatial point, the processor is further configured to:

Acquiring sampling results within the historical perception range of the depth sensor;

Estimating a reliable area within the current perception range according to the sampling result;

The spatial points are sampled within the reliable region.
The device according to claim 52, wherein the processor is further configured to: sample space points at a high sampling rate in the reliable region, and sample at a low sampling rate in the unreliable region of the current sensing range Rate sampling space point.
The device according to claim 52, wherein the reliable region is determined according to a safe channel generated from the sampling result.
The device according to claim 52, wherein the historical sensing range and the current sensing range have an overlapping area.
The device according to claim 52, wherein after sampling the spatial point in the reliable region, the processor is further configured to:

determining the distance between the spatial point and the nearest neighbor obstacle according to the obstacle information obtained within the current sensing range and/or the historical sensing range;

Spatial points whose distance is greater than a preset distance threshold are reserved.
The device according to claim 51, wherein the spatial points are added to the path search tree in descending order of distance from the root node.
The device according to claim 49, wherein after said determining a path search tree extending to said termination track point, said processor is further configured to:

According to the relationship between each leaf node and the nearest neighbor obstacle in the path search tree, determine the safety factor of each leaf node;

Traversing the path search tree from each leaf node to determine the safety factor of each non-leaf node;

The path search tree is pruned according to the safety factor of each node in the path search tree.
The device according to claim 58, wherein the safety factor of the non-leaf node is the sum of the safety factors of its child nodes.
The apparatus of claim 58, wherein the processor is further configured to:

For each leaf node, determine the direction of the leaf node towards the nearest neighbor obstacle;

The safety factor of the leaf node is determined according to the angle between the direction and the growth direction of the leaf node.
The device according to claim 60, wherein the safety factor of the leaf node is positively correlated with the included angle.
The device according to claim 60, wherein if the included angle is greater than a preset angle threshold, the safety factor of the leaf node is a first preset value; otherwise, the safety factor of the leaf node is a second preset value. A preset value, the first preset value is greater than the second preset value.
The device according to claim 60, wherein the growth direction of the leaf node is the direction of a connection line between the leaf node and its parent node.
The device according to claim 58, wherein the processor is further configured to: for each leaf node, determine the safety of the leaf node according to the distance between the leaf node and the nearest neighbor obstacle. coefficient.
The device according to claim 64, wherein the safety factor of the leaf node is positively correlated with the distance.
The device according to claim 58, wherein the processor is further configured to:

The safety factor of each node in the path search tree is normalized.
The device according to claim 66, wherein the normalization process comprises: for each node, determining the ratio between its own safety factor and the maximum safety factor among nodes belonging to the same parent node as the normalized Normalized value.
The device according to any one of claims 58 to 67, wherein the pruning process includes: pruning nodes in the path search tree whose safety factor is lower than a preset threshold.
The device according to claim 58, wherein the nearest neighbor obstacle is obtained from pre-stored obstacle information based on the position of the leaf node; the obstacle information is obtained by the depth sensor in its current perception range and/or within historical perception.
The apparatus according to claim 37, wherein the secure channel is composed of a plurality of secure regions generated in the secure path.
The device according to claim 70, wherein the center points of the plurality of safe areas are all in the safe path;

The center point of one of the security areas is the target leaf node, and the center points of the other security areas are the intersection points of the previous security area and the security path.
The device according to claim 70 or 71, wherein the size of each safety area is determined according to the distance between a part of the safety path corresponding to the safety area and the nearest obstacle.
A mobile platform, characterized in that it comprises:

body;

a power system, arranged in the fuselage, for driving the movement of the movable platform;

And, the trajectory generating device according to any one of claims 37 to 72.
A computer-readable storage medium, wherein the computer-readable storage medium stores executable instructions, and when the executable instructions are executed by a processor, the method according to any one of claims 1 to 36 is implemented.