Non-navigation exploration planning method and system based on spatio-temporal combination

Abstract

The invention provides a navigation-free exploration planning method and system based on spatio-temporal combination. The navigation-free exploration planning method comprises the following steps: acquiring a basic spatio-temporal grid map; projecting the prediction information of the dynamic obstacle to a basic space-time grid map based on a time alignment condition to construct a target space-time grid map; based on the target space-time grid map and the vehicle collision condition, obtaining a track set of the target vehicle according to the speed domain and the angular speed domain; and performing cost calculation on the i first trajectories in the trajectory set to obtain an optimal trajectory. The technical problem of low planning efficiency of the vehicle in path planning due to the fact that three-layer traversal needs to be carried out in collision calculation of the existing related algorithm is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of safe operation of vehicles, and in particular to a navigation-free exploration planning method based on time-space combination and a navigation-free exploration planning system based on time-space combination. Background Art

[0002] During the vehicle driving process, it is necessary to perceive, analyze and calculate the environmental information, combine the vehicle's own positioning and target position, and use obstacle avoidance and vehicle path planning methods to achieve autonomous navigation. Most of the existing collision algorithms traverse obstacles to make corresponding collisions, O (number of obstacles, obstacle trajectory, vehicle trajectory), three-layer traversal, and complex algorithms will directly maximize the chip computing power, and most of the time it is not enough to cope with the corresponding situation.

[0003] Therefore, there is at least one of the following problems in the related technology: the existing related algorithms need to perform three-layer traversal when calculating collisions, resulting in low planning efficiency in vehicle path planning. Summary of the invention

[0004] The present invention solves the technical problem that the existing related algorithms need to perform three-layer traversal during collision calculation, resulting in low planning efficiency in vehicle path planning.

[0005] To solve the above problems, the present invention provides a navigation-free exploration planning method based on space-time union, comprising: obtaining a basic space-time grid map; based on a time alignment condition, projecting the prediction information of dynamic obstacles onto the basic space-time grid map to construct a target space-time grid map; based on the target space-time grid map and vehicle collision conditions, obtaining a trajectory set of the target vehicle according to the velocity domain and the angular velocity domain; and performing cost calculation on i first trajectories in the trajectory set to obtain the optimal trajectory.

[0006] Compared with the prior art, the technical effect achieved by adopting this technical solution is as follows: In this application, the prediction information can also be called dynamic information, that is, the predicted trajectory of dynamic obstacles. Based on the time alignment principle, the perceived objects (mainly including static obstacles) are constructed into corresponding inaccessible areas to form a basic space-time grid map; at the same time, the dynamic obstacles are projected onto relevant time slices according to their predicted trajectories to construct inaccessible areas. In this way, the target space-time grid map is constructed. The planning scheme combined with space-time information can effectively deal with the coordinate conversion problem caused by space-time separation planning, and the target space-time grid map is constructed through the space-time layering method, thereby improving the planning efficiency.

[0007] At the same time, the present application has a strong ability to handle dynamic obstacles. Most of the current algorithms traverse obstacles to make corresponding collisions, O (number of obstacles, obstacle trajectories, vehicle trajectories), three-layer traversal, will directly max out the chip computing power, and most of the time it is not enough to cope with the corresponding situation. However, the present application directly reduces the algorithm complexity to O (vehicle trajectory) by constructing the corresponding target spatiotemporal grid map, reducing it from O(n^3) to O(n), which is a geometric level algorithm performance optimization; the target spatiotemporal grid map is applicable to almost all existing planning algorithms, which will make the navigation-free exploration planning method of the present application applicable to almost all low-speed scenarios, including but not limited to forward exploration, narrow road forward, backward exploration, and tracking and reversing.

[0008] In one example of the present invention, based on the time alignment condition, the predicted information of the dynamic obstacles is projected onto the basic space-time grid map to construct the target space-time grid map, including: obtaining the dynamic set of the dynamic obstacles under the basic space-time grid map; stacking the dynamic set on the corresponding basic space-time grid map in time order to construct the target space-time grid map; wherein the dynamic set includes: the predicted position of the dynamic obstacle at each moment.

[0009] Compared with the existing technology, the technical effect achieved by adopting this technical solution is: the real world is described by constructing a target spatiotemporal grid map, without the need to conduct in-depth analysis of the attributes of each object, that is, static cars and trucks are equivalent, and both are inaccessible environmental areas, rather than the existing algorithms that require them to be accurately distinguished before they can be used, and the computing efficiency is greatly improved. At the same time, it can also make better compatibility with the way the existing algorithms give object-related attributes, that is, convert them into inaccessible areas of the spatiotemporal grid map. In this way, it can support both low-perception and high-perception platforms.

[0010] In one example of the present invention, the target spatiotemporal grid map and the vehicle collision condition are based on the speed domain and the angular velocity domain to obtain the trajectory set of the target vehicle, including: obtaining j first trajectories of the target vehicle according to the speed domain and the angular velocity domain; performing a coarse collision calculation on the j first trajectories; performing a fine collision calculation on the first trajectories that pass the coarse collision calculation; and storing the i first trajectories that pass the fine collision calculation in the trajectory set, i≤j.

[0011] In one example of the present invention, the first trajectory includes k trajectory points; the rough collision calculation of the j first trajectories includes: obtaining an index range of the target vehicle at each of the trajectory points according to the first trajectory; based on the index range, determining whether there are occupied trajectory points on the first trajectory; when there are occupied trajectory points on the first trajectory, determining that the first trajectory has not passed the rough collision calculation.

[0012] Compared with the prior art, the technical effects achieved by adopting this technical solution are as follows: giving priority to rough collision calculation and then fine collision calculation, which can reduce the amount of calculation in the collision calculation process and improve the accuracy and efficiency of the collision calculation; compared with the obstacle monitoring of the existing collision algorithm which only monitors static obstacles or traverses dynamic obstacles to perform collisions over and over again, this collision algorithm only needs to extract the target space-time grid map of the corresponding time slice and perform a collision detection once to end the calculation of the collision detection part.

[0013] In one embodiment of the present invention, the trajectory set is: , the cost calculation of the i first trajectories in the trajectory set to obtain the optimal trajectory includes: cost calculation of the first trajectories using a cost function; the cost function is: ;in, is the trajectory set, is the i-th first trajectory in the trajectory set; For the trajectory The speed cost function is For the trajectory The obstacle distance cost function is For the trajectory The angular velocity cost function is: ; The obstacle distance cost function is: ; The angular velocity cost function is: ;in, is the speed cost coefficient, For the trajectory At the track point speed, The maximum speed limited by the scene; is the obstacle distance cost coefficient, It's a track At the track point Distance to obstacles; is the angular velocity cost coefficient, For the trajectory At the track point The angular velocity of The maximum angular velocity limited by the scene.

[0014] Compared with the prior art, the technical effect achieved by adopting this technical solution is as follows: traverse the i first trajectories in the trajectory set, and for each trajectory point of each first trajectory Calculate the cost of each first trajectory by calculating the speed cost, obstacle distance cost and angular velocity cost: , is the kth track point in the i-th first track. According to the cost calculation of the i-th first track, the corresponding set is obtained: ,in, is the cost of the 0th first trajectory; is the cost of the first trajectory of the i-th item. Finally, the first trajectory with the highest cost function is obtained and published to the downstream control as the optimal trajectory for vehicle control.

[0015] In one example of the present invention, before obtaining the basic space-time grid map, the non-navigation exploration planning method also includes: obtaining upstream information, the upstream information including: positioning data, perception data, prediction data and chassis data; based on the positioning data, time-aligning the upstream information to obtain time synchronization data; constructing the environmental model according to the time synchronization data; and constructing the basic space-time grid map according to the environmental model.

[0016] Compared with the existing technology, the technical effect achieved by adopting this technical solution is as follows: this application only needs simple corner radar and ultrasonic radar, and knows the corresponding obstacle position and height in the real world, so as to create an abstract environment for the real world. At the same time, a simple single-vision camera is used to perform simple identification and tracking of dynamic obstacles, so as to deal with static obstacles and dynamic obstacles at the same time; while effectively reducing development costs and deployment costs, the performance can be on par with existing algorithms.

[0017] In one example of the present invention, the speed domain and angular velocity domain of the target vehicle are obtained based on the scene restrictions and the vehicle's own parameter restrictions, including: obtaining a vehicle-related speed interval, an acceleration-related speed interval, and a collision-related speed interval; obtaining a vehicle-related angular velocity interval, an acceleration-related angular velocity interval, and a collision-related angular velocity interval; obtaining a speed domain according to the vehicle-related speed interval, the acceleration-related speed interval, and the collision-related speed interval; obtaining an angular velocity domain according to the vehicle-related angular velocity interval, the acceleration-related angular velocity interval, and the collision-related angular velocity interval; wherein the vehicle-related speed interval is: , the acceleration-related speed range is: , the collision-related velocity range is: ; The vehicle-related angular velocity range is: , the acceleration-related angular velocity range is: , the collision-related angular velocity range is: .

[0018] Compared with the existing technology, the technical effect achieved by adopting this technical solution is: based on the scene restrictions and the vehicle's own parameter restrictions, the vehicle-related speed range of the target vehicle is obtained. ; Obtain the acceleration-related speed range of the target vehicle based on comfort limits Angular velocity range related to acceleration ; Obtain the collision-related speed range of the target vehicle based on the collision limit Angular velocity range related to collision .

[0019] In one embodiment of the present invention, the velocity domain is: ; The angular velocity domain is: ;in, is the first vehicle-related speed threshold, is the first acceleration-related speed threshold, is the first collision-related speed threshold; is the second vehicle-related speed threshold, is the second acceleration-related speed threshold, is the second collision-related speed threshold; is the first vehicle-related angular velocity threshold, is the first acceleration-related angular velocity threshold, is the first collision-related angular velocity threshold; is the second vehicle-related angular velocity threshold, is the second acceleration-related angular velocity threshold, is the second collision-related angular velocity threshold.

[0020] Compared with the prior art, the technical effect achieved by adopting the technical solution is as follows: after obtaining the vehicle-related speed interval, the acceleration-related speed interval, the collision-related speed interval and the vehicle-related angular velocity interval, the acceleration-related angular velocity interval and the collision-related angular velocity interval, the speed domain can be obtained through the vehicle-related speed interval, the acceleration-related speed interval and the collision-related speed interval; the angular velocity domain can be obtained through the vehicle-related angular velocity interval, the acceleration-related angular velocity interval and the collision-related angular velocity interval.

[0021] On the other hand, an embodiment of the present invention also provides a non-navigation exploration planning system based on time-space union, using a non-navigation exploration planning method based on time-space union as in any one of the above embodiments, including: a first acquisition module, the first acquisition module is used to acquire a basic space-time grid map; a construction module, the construction module is used to project the prediction information of dynamic obstacles onto the basic space-time grid map based on a time alignment condition to construct a target space-time grid map; a second acquisition module, the second acquisition module is used to acquire a trajectory set of the target vehicle according to the speed domain and the angular velocity domain based on the target space-time grid map and the vehicle collision condition; a calculation module, the calculation module is used to perform cost calculation on i first trajectories in the trajectory set to obtain the optimal trajectory.

[0022] Compared with the prior art, the technical effect achieved by adopting this technical solution is as follows: the non-navigation exploration planning system based on time-space union in this embodiment is used to implement the non-navigation exploration planning method based on time-space union as in any embodiment of the present invention, so it has all the beneficial effects of the non-navigation exploration planning method based on time-space union as in any embodiment of the present invention, which will not be repeated here.

[0023] After adopting the technical solution of the present invention, the following technical effects can be achieved: (1) The present application directly reduces the algorithm complexity to O(vehicle trajectory) by constructing the corresponding target spatiotemporal grid map, reducing it from O(n^3) to O(n), which is a geometric level algorithm performance optimization; the target spatiotemporal grid map is applicable to almost all existing planning algorithms, which makes the navigation-free exploration planning method of the present application applicable to almost all low-speed scenarios, including but not limited to forward exploration, narrow road forward, backward exploration, and tracking reverse; (2) Based on the time alignment principle, the sensed objects (mainly static obstacles) are mapped to corresponding inaccessible areas to form a basic space-time grid map. At the same time, dynamic obstacles are projected onto relevant time slices according to their predicted trajectories to construct inaccessible areas. In this way, the target space-time grid map is constructed. The planning scheme combined with space-time information can effectively deal with the coordinate conversion problem caused by space-time separation planning. The target space-time grid map is constructed through the space-time stratification method, thereby improving planning efficiency. (3) This application only requires simple corner radar and ultrasonic radar. By knowing the direction and height of the corresponding obstacles in the real world, it can create an abstract environment for the real world. At the same time, a simple single-vision camera can be used to perform simple identification and tracking of dynamic obstacles, so that static obstacles and dynamic obstacles can be dealt with at the same time. While effectively reducing development and deployment costs, the performance can be on par with existing algorithms. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings to be used in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work. Figure 1 A flowchart of a non-navigation exploration planning method based on spatiotemporal union provided in Example 1 of the present invention; Figure 2 A specific flow chart of constructing a target spatiotemporal grid map provided in the first embodiment of the present invention; Figure 3 A schematic structural diagram of a non-navigation exploration planning system based on spatiotemporal union provided in the second embodiment of the present invention.

[0025] Description of reference numerals: 100 - non-navigation exploration planning system; 101 - first acquisition module; 102 - construction module; 103 - second acquisition module; 104 - calculation module. DETAILED DESCRIPTION

[0026] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and understandable, the technical solutions in the embodiments of the present application are described clearly and completely. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application.

[0027] As mentioned above, the existing related algorithms need to perform three-layer traversal when calculating collisions, which results in low planning efficiency in vehicle path planning and inability to achieve high-efficiency planning.

[0028] In view of this, the present application provides a non-navigation exploration planning method and a non-navigation exploration planning system based on spatiotemporal union, in order to solve the above technical problems. The scheme of the present application will be described below in combination with multiple embodiments. It should be noted that similar or related contents in the following embodiments can be referenced to each other and will not be repeated later.

[0029] [Example 1] See also Figure 1 , which is a flowchart of a non-navigation exploration planning method based on spatiotemporal union provided by an embodiment of the present invention, combined with Figure 2 , non-navigation exploration planning methods include: S100: Obtaining a basic spatiotemporal grid map; S200: Based on the time alignment condition, the prediction information of the dynamic obstacle is projected onto the basic space-time grid map to construct a target space-time grid map; S300: Based on the target spatiotemporal grid map and the vehicle collision condition, a trajectory set of the target vehicle is obtained according to the velocity domain and the angular velocity domain; S400: Calculate the cost of i first trajectories in the trajectory set to obtain an optimal trajectory.

[0030] Combination Figure 2 In a specific embodiment, the basic spatiotemporal grid map is a map containing static obstacles (such as buildings, green belts, stationary vehicles, etc.) in the current environment. Due to the development of society, there are updates to the current environment, so the basic spatiotemporal grid map will change with the static obstacles in the current environment. For example, Figure 2 The "black vehicle logo" shown in the figure is used to represent static obstacles, the "pink vehicle logo" is used to represent dynamic obstacles, and the "blue vehicle logo" is used to represent the vehicle you are driving; the "pedestrian logo" in the figure is used to represent pedestrians, and the "..." is used to represent the predicted trajectory of dynamic targets.

[0031] In this application, the prediction information can also be called dynamic information, that is, the predicted trajectory of dynamic obstacles. Based on the principle of time alignment, the perceived objects (mainly including static obstacles) are constructed into corresponding inaccessible areas to form a basic space-time grid map; at the same time, the dynamic obstacles are projected onto the relevant time slices according to their predicted trajectories to construct inaccessible areas. In this way, the target space-time grid map is constructed. The planning scheme combined with the space-time information can effectively deal with the coordinate conversion problem caused by the space-time separation planning, and the target space-time grid map is constructed through the space-time layering method, thereby improving the planning efficiency.

[0032] At the same time, the present application has a strong ability to handle dynamic obstacles. Most of the current algorithms traverse obstacles to make corresponding collisions, O (number of obstacles, obstacle trajectories, ego vehicle trajectories), three-layer traversal, will directly max out the chip computing power, and most of the time it is not enough to cope with the corresponding situation. However, the present application directly reduces the algorithm complexity to O (ego vehicle trajectory) by constructing the corresponding target spatiotemporal grid map, reducing it from O(n^3) to O(n), which is a geometric level algorithm performance optimization; the target spatiotemporal grid map is applicable to almost all existing planning algorithms, which will make the navigation-free exploration planning method of the present application applicable to almost all low-speed scenarios, including but not limited to forward exploration, narrow road forward, backward exploration, and tracking and reversing. Among them, the trajectory of the ego vehicle is the trajectory of the target vehicle.

[0033] Furthermore, S200 includes: S210: Obtain a dynamic set of dynamic obstacles under the basic space-time grid map; S220: stacking the dynamic set onto the corresponding basic spatiotemporal grid map in chronological order to construct a target spatiotemporal grid map; The dynamic set includes: the predicted positions of dynamic obstacles at each moment.

[0034] Specifically, for static obstacles, the main task is to construct the basic occupied area, and the basic space-time grid map is: , For The base grid map at the moment, For The basic grid map at the moment. Among them, express The timestamp corresponding to the data item, where the larger the number, the longer it is from the real time. is the latest message received, and n represents the length of the message queue.

[0035] Dynamic collection is , It represents the prediction information of all dynamic obstacles under the basic space-time grid map, that is, the dynamic set.

[0036] The target spatiotemporal grid map includes (n+1) target grid maps. , Indicated in The target grid map at the moment; , Indicated in The position of the (m+1) dynamic obstacles at the moment, For the first dynamic obstacle The location at the moment, is the (m+1)th dynamic obstacle at The location at the moment. By constructing a target spatiotemporal grid map to describe the real world, there is no need to conduct in-depth analysis of the attributes of each object, that is, static cars and trucks are equivalent and are both inaccessible environmental areas, rather than the existing algorithms that require them to be accurately distinguished before they can be used, which greatly improves the computing efficiency. At the same time, it can also make better compatibility with the existing algorithms for giving object-related attributes, that is, converting them into inaccessible areas of the spatiotemporal grid map. In this way, it can support both low-perception and high-perception platforms.

[0037] Furthermore, S300 includes: S310: Obtain j first trajectories of the target vehicle according to the velocity domain and the angular velocity domain; S320: performing rough collision calculation on the j first trajectories; S330: performing fine collision calculation on the first trajectory through the coarse collision calculation; S340: Store i first trajectories calculated through fine collision into a trajectory set, i≤j.

[0038] Specifically, the rough collision calculation is performed first and then the fine collision calculation is performed, which can reduce the amount of calculation in the collision calculation process and improve the accuracy and efficiency of the collision calculation. Compared with the obstacle monitoring of the existing collision algorithm, which only monitors static obstacles or traverses dynamic obstacles to perform collisions over and over again, this collision algorithm only needs to extract the target space-time grid map of the corresponding time slice and perform a collision detection once to end the calculation of the collision detection part.

[0039] Further, the first trajectory includes k trajectory points, and S320 includes: S321: Obtaining the index range of the target vehicle at each trajectory point according to the first trajectory; S322: Based on the index range, determine whether there is an occupied track point on the first track; S323: When there is an occupied trajectory point on the first trajectory, it is determined that the first trajectory does not pass the rough collision calculation.

[0040] Specifically, a collision calculation is performed on each first trajectory of the target vehicle (that is, j first trajectories), and specifically, a rough collision calculation is performed on each first trajectory first and then a fine collision calculation is performed. The specific method of the rough collision calculation is: traverse each trajectory point of the first trajectory to find whether there is an occupied trajectory point on the first trajectory; specifically: For in , for The target vehicle rear axle center position at time Target grid map at the moment and the trajectory points of the target vehicle, One-to-one correspondence with the trajectory points of the target vehicle; calculate the four corners of the target vehicle The first index value of: , calculate the horizontal axis index value and the vertical axis index value of the target vehicle at this time according to the first index value: , , , , get the index range according to the horizontal axis index value and the vertical axis index value: ; Perform rough collision calculation based on index range: From to From to If Occupied: Detailed collision calculation.

[0041] in, , , , Represents the four corners of the target vehicle The four coordinates below; , , , Respectively represent the four corners that constitute the index range The four coordinates below; The present invention performs rough collision calculation based on the index range, rather than based on the target grid map, thereby narrowing the range of collision calculation and reducing the amount of calculation.

[0042] Preferably, the specific method of fine collision calculation is: ①. Convert to the vehicle body coordinate system; ②. For the horizontal axis (i.e. the direction of the front of the target vehicle), determine ③. If the horizontal axis and the vertical axis are both within the boundary of the target vehicle, it means that the target vehicle has not passed the detailed collision calculation.

[0043] Furthermore, the trajectory set is: , S300 includes: Calculate the cost of the first trajectory through the cost function; The cost function is: ;in, is a set of trajectories, is the i-th first trajectory in the trajectory set; For the trajectory The speed cost function is For the trajectory The obstacle distance cost function is For the trajectory Angular velocity cost function; The speed cost function is: ; The obstacle distance cost function is: ; The angular velocity cost function is: ; in, is the speed cost coefficient, For the trajectory At the track point speed, The maximum speed limited by the scene; is the obstacle distance cost coefficient, It's a track At the track point Distance to obstacles; is the angular velocity cost coefficient, For the trajectory At the track point The angular velocity of The maximum angular velocity limited by the scene.

[0044] Specifically, traverse the i first trajectories in the trajectory set, and for each trajectory point of each first trajectory Calculate the cost of each first trajectory by calculating the speed cost, obstacle distance cost and angular velocity cost: , is the kth track point in the i-th first track. According to the cost calculation of the i-th first track, the corresponding set is obtained: ,in, is the cost of the 0th first trajectory; is the cost of the first trajectory of the i-th item. Finally, the first trajectory with the highest cost function is obtained and published to the downstream control as the optimal trajectory for vehicle control.

[0045] Preferably, ;when hour, = ;when hour, = .

[0046] Furthermore, before S100, the non-navigation exploration planning method also includes: Obtain upstream information, including positioning data, perception data, prediction data, and chassis data; Based on the positioning data, the upstream information is time-aligned to obtain time-synchronized data; Constructing an environmental model based on time-synchronized data; Construct a basic space-time grid map based on the environmental model.

[0047] Specifically, upstream information is generally obtained through weak perception and strong perception. Weak perception only has relatively simple sensors such as fisheye, single camera, and ultrasonic wave. In this solution, freespace and some visible obstacles are predicted. Strong perception uses more powerful sensors, including high-precision sensors such as LiDAR to detect real-world information. In this solution, lane lines, obstacles, obstacle prediction trajectories, etc. are obtained. Upstream information includes but is not limited to: positioning data, perception data, prediction data, and chassis data.

[0048] For the received upstream information, time alignment is performed based on the positioning data, that is, the message with the closest timestamp to the positioning data is obtained as the information to be fused: , , , , , where: , , , , It is the timestamp corresponding to each message in each upstream message queue; As the benchmark time, traverse , , , , respectively , , , Get from The data corresponding to the most recent time is taken as the time synchronization data and recorded as , , , , Go to the next step to construct the corresponding environment model. Among them, localization is positioning data, perception is perception data, prediction is prediction data, and canbus is chassis data; , , , They represent positioning data messages, perception data messages, prediction data messages and chassis data messages respectively; MSG is a data message, which can also be represented as time synchronization data.

[0049] by For example, Indicates that the predicted data is Time synchronization data at a certain moment can also be expressed as Prediction data message at the moment; Indicates that the predicted data is Time synchronization data at a certain moment can also be expressed as Forecast data message at the moment.

[0050] This application only needs simple corner radar and ultrasonic radar, and knows the direction and height of the corresponding obstacles in the real world, so as to create an abstract environment for the real world. At the same time, a simple single-vision camera is used to perform simple identification and tracking of dynamic obstacles, so as to deal with static obstacles and dynamic obstacles at the same time. While effectively reducing development and deployment costs, the performance can be on par with existing algorithms.

[0051] Furthermore, before S300, the non-navigation exploration planning method further includes: Obtain vehicle-related speed intervals, acceleration-related speed intervals, and collision-related speed intervals; Obtain vehicle-related angular velocity intervals, acceleration-related angular velocity intervals, and collision-related angular velocity intervals; A speed domain is obtained according to a vehicle-related speed interval, an acceleration-related speed interval, and a collision-related speed interval; Obtaining an angular velocity domain according to a vehicle-related angular velocity interval, an acceleration-related angular velocity interval, and a collision-related angular velocity interval; Among them, the vehicle-related speed range is: , the acceleration-related speed range is: , the collision-related velocity range is: ; The vehicle-related angular velocity range is: , the acceleration-related angular velocity range is: , the collision-related angular velocity range is: .

[0052] Specifically, the vehicle-related speed range of the target vehicle is obtained based on the scene restrictions and the vehicle's own parameter restrictions. The parking lot itself has a maximum speed constraint, and its driving principles need to be followed: , is the first vehicle-related speed threshold, is the second vehicle-related speed threshold; preferably, ; , which is the maximum speed limited by the scene. Based on the steering wheel limit of the vehicle itself, the vehicle-related angular velocity range of the target vehicle is obtained , the steering wheel of the target vehicle itself has a limit, which can be defined as: , is the first vehicle-related angular velocity threshold, is the second vehicle-related angular velocity threshold; preferably, , which is the minimum angular velocity value related to the vehicle body; , which is the maximum angular velocity value related to the vehicle body.

[0053] Obtain the acceleration-related speed range of the target vehicle based on comfort constraints Angular velocity range related to acceleration In a parking lot environment, both acceleration and deceleration need to be constrained. Rapid acceleration is not allowed, otherwise it will cause unsafe impacts on other people and other cars. Angular acceleration refers to the rate of change of the driver's steering wheel. The steering wheel will not change significantly per unit time, otherwise it will cause discomfort and may even cause harm to the vehicle and other cars: , is the first acceleration-related speed threshold, is the second acceleration-related speed threshold; preferably, , , is the speed of the target vehicle at the beginning of planning, is the difference in planning time points, is the minimum acceleration in the current scene, It is the maximum acceleration value in the current scene. , is the first acceleration-related angular velocity threshold, is the second acceleration-related angular velocity threshold; preferably, , , is the angular velocity of the ego vehicle at the beginning of planning; is the difference in planning time points; is the minimum angular acceleration in the current scene; is the maximum angular acceleration in the current scene.

[0054] Obtain collision-related speed range of target vehicle based on collision limit Angular velocity range related to collision , obtain the nearest collision distance of the target vehicle in the current state, and calculate the maximum speed that can be braked to ensure a safe distance: , is the first collision-related velocity threshold, is the second collision-related speed threshold; preferably, , , is the Euclidean distance between the target vehicle and the obstacle at the current relative speed and angular velocity. , is the first collision-related angular velocity threshold, is the second collision-related angular velocity threshold; preferably, , , for It is the maximum angular velocity in the current scene.

[0055] This application performs path planning through the vehicle's own parameters, scenario restrictions and driving parameters to minimize the demand for hardware. It can be deployed on a platform with lower computing power, without the need for large computing power support like deep learning algorithms. The path planning meets the requirements of vehicle dynamics and is in line with the physical laws of the real world. It is also a complete white-box planning that can effectively track and solve problems, rather than a black-box strategy like deep learning.

[0056] Furthermore, the velocity domain is: ; The angular velocity domain is: ;in, is the first vehicle-related speed threshold, is the first acceleration-related speed threshold, is the first collision-related speed threshold; is the second vehicle-related speed threshold, is the second acceleration-related speed threshold, is the second collision-related speed threshold; is the first vehicle-related angular velocity threshold, is the first acceleration-related angular velocity threshold, is the first collision-related angular velocity threshold; is the second vehicle-related angular velocity threshold, is the second acceleration-related angular velocity threshold, is the second collision-related angular velocity threshold.

[0057] Specifically, after obtaining the vehicle-related speed interval, acceleration-related speed interval, collision-related speed interval and the vehicle-related angular velocity interval, acceleration-related angular velocity interval, and collision-related angular velocity interval, the speed domain can be obtained through the vehicle-related speed interval, acceleration-related speed interval, and collision-related speed interval; the angular velocity domain can be obtained through the vehicle-related angular velocity interval, acceleration-related angular velocity interval, and collision-related angular velocity interval.

[0058] Among them, it can be expressed as , , is the first speed threshold, is the second speed threshold; , , is the first angular velocity threshold, is the second angular velocity threshold.

[0059] [Example 2] See also Figure 3The present embodiment also provides a non-navigation exploration planning system 100 based on time-space union, using the non-navigation exploration planning method as in the first embodiment, the non-navigation exploration planning system 100 includes: a first acquisition module 101, a construction module 102, a second acquisition module 103 and a calculation module 104, the first acquisition module 101 is used to acquire a basic space-time grid map; the construction module 102 is used to project the prediction information of the dynamic obstacle onto the basic space-time grid map based on the time alignment condition to construct a target space-time grid map; the second acquisition module 103 is used to acquire a trajectory set of the target vehicle according to the speed domain and the angular velocity domain based on the target space-time grid map and the vehicle collision condition; the calculation module 104 is used to perform cost calculation on i first trajectories in the trajectory set to obtain the optimal trajectory.

[0060] In a specific embodiment, the first acquisition module 101, the construction module 102, the second acquisition module 103 and the calculation module 104 of the non-navigation exploration planning system 100 cooperate to implement the non-navigation exploration planning method based on time-space union as described in the first embodiment above, which will not be repeated here.

[0061] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A navigation-free exploration planning method based on spatiotemporal union, characterized in that: The non-navigation exploration planning method comprises: Get the basic space-time grid map; Based on the time alignment condition, the prediction information of the dynamic obstacle is projected onto the basic space-time grid map to construct a target space-time grid map; Based on the target spatiotemporal grid map and the vehicle collision condition, a trajectory set of the target vehicle is obtained according to the velocity domain and the angular velocity domain; Cost calculation is performed on i first trajectories in the trajectory set to obtain an optimal trajectory.

2. The non-navigation exploration planning method according to claim 1, characterized in that: The method of projecting the prediction information of the dynamic obstacles onto the basic spatiotemporal grid map based on the time alignment condition to construct the target spatiotemporal grid map includes: Obtain a dynamic set of the dynamic obstacles under the basic space-time grid map; Stacking the dynamic set onto the corresponding basic space-time grid map in chronological order to construct the target space-time grid map; The dynamic set includes: the predicted positions of the dynamic obstacles at various moments.

3. The non-navigation exploration planning method according to claim 1, characterized in that: The step of acquiring a trajectory set of a target vehicle based on the target spatiotemporal grid map and the vehicle collision condition according to the velocity domain and the angular velocity domain comprises: Acquire j first trajectories of the target vehicle according to the velocity domain and the angular velocity domain; Performing rough collision calculation on the j first trajectories; Performing fine collision calculation on the first trajectory obtained through coarse collision calculation; The i first trajectories calculated through fine collision are stored in the trajectory set, i≤j.

4. The non-navigation exploration planning method according to claim 3, characterized in that: The first trajectory includes k trajectory points; The performing rough collision calculation on the j first trajectories includes: Acquire the index range of the target vehicle at each of the trajectory points according to the first trajectory; Based on the index range, determining whether there is an occupied track point on the first track; When there are occupied track points on the first track, it is determined that the first track does not pass the rough collision calculation.

5. The non-navigation exploration planning method according to claim 1, wherein the trajectory set is: , characterized in that, The performing cost calculation on the i first trajectories in the trajectory set to obtain the optimal trajectory includes: Calculating the cost of the first trajectory using a cost function; The cost function is: ; in, is the trajectory set, is the i-th first trajectory in the trajectory set; For the trajectory The speed cost function is For the trajectory The obstacle distance cost function is For the trajectory Angular velocity cost function; The speed cost function is: ; The obstacle distance cost function is: ; The angular velocity cost function is: ; in, is the speed cost coefficient, For the trajectory At the track point speed, The maximum speed limited by the scene; is the obstacle distance cost coefficient, It's a track At the track point distance to obstacles; is the angular velocity cost coefficient, For the trajectory At the track point The angular velocity of The maximum angular velocity limited by the scene.

6. The non-navigation exploration planning method according to claim 1, characterized in that: Before acquiring the basic spatiotemporal grid map, the non-navigation exploration planning method further includes: Acquire upstream information, the upstream information including: positioning data, perception data, prediction data and chassis data; Based on the positioning data, time aligning the upstream information to obtain time synchronization data; constructing an environment model based on the time synchronization data; The basic spatiotemporal grid map is constructed according to the environmental model.

7. The non-navigation exploration planning method according to claim 1, characterized in that: Before acquiring the trajectory set of the target vehicle according to the speed domain and the angular velocity domain based on the target spatiotemporal grid map and the vehicle collision condition, the non-navigation exploration planning method further includes: Obtain vehicle-related speed intervals, acceleration-related speed intervals, and collision-related speed intervals; Obtain vehicle-related angular velocity intervals, acceleration-related angular velocity intervals, and collision-related angular velocity intervals; Obtaining the speed domain according to the vehicle-related speed interval, the acceleration-related speed interval, and the collision-related speed interval; Obtaining the angular velocity domain according to the vehicle-related angular velocity interval, the acceleration-related angular velocity interval, and the collision-related angular velocity interval; Wherein, the vehicle-related speed range is: , the acceleration-related speed range is: , the collision-related speed range is: ; The vehicle-related angular velocity range is: , the acceleration-related angular velocity interval is: , the collision-related angular velocity interval is: .

8. The non-navigation exploration planning method according to claim 7, characterized in that: The speed domain is: ; The angular velocity domain is: ; in, is the first vehicle-related speed threshold, is the first acceleration-related speed threshold, is the first collision-related speed threshold; is the second vehicle-related speed threshold, is the second acceleration-related speed threshold, is the second collision-related speed threshold; is the first vehicle-related angular velocity threshold, is the first acceleration-related angular velocity threshold, is the first collision-related angular velocity threshold; is the second vehicle-related angular velocity threshold, is the second acceleration-related angular velocity threshold, is the second collision-related angular velocity threshold.

9. A non-navigation exploration planning system based on time-space combination, wherein the non-navigation exploration planning system based on time-space combination uses the non-navigation exploration planning method according to any one of claims 1 to 8, characterized in that: include: A first acquisition module, wherein the first acquisition module is used to acquire a basic spatiotemporal grid map; A construction module, the construction module is used to project the prediction information of the dynamic obstacle onto the basic spatiotemporal grid map based on the time alignment condition to construct a target spatiotemporal grid map; A second acquisition module, the second acquisition module is used to acquire a trajectory set of the target vehicle according to a velocity domain and an angular velocity domain based on the target spatiotemporal grid map and the vehicle collision condition; A calculation module is used to perform cost calculation on i first trajectories in the trajectory set to obtain an optimal trajectory.

Images