Multi-Model Based Complex Scene Path Planning Generation Method and System

Abstract

The present invention discloses a method and system for generating complex-scenario path planning based on multiple models, which relates to the technical field of path planning. The static environment information set and the dynamic environment information set in the road are collected in advance. Based on the static environment information set and the dynamic environment information set, a multi-layer feature representation of the road is constructed. Based on the multi-layer feature representation of the road, a path planning model is constructed. Based on the path planning model, an optimization model is used to generate a planned path. It can significantly improve the path planning ability of autonomous vehicles in complex scenarios.

Description

Technical Field

[0001] The present invention relates to the technical field of path planning, and specifically to a method and system for generating complex-scene path planning based on multiple models. Background Art

[0002] Path planning in complex scenes has always been a major challenge in the field of autonomous driving. Most traditional path planning methods are based on simple environmental models and are difficult to handle complex actual road scenes. With the continuous development of autonomous driving technology, the requirements for path planning algorithms are also increasing day by day, and various complex factors such as static obstacles, dynamic obstacles, and traffic rules need to be considered.

[0003] In the prior art, some methods only consider static obstacles and ignore the influence of dynamic obstacles, which may lead to potential collision risks in the planned path. Other methods, although considering dynamic obstacles, fail to make full use of global topological information and local topological features, and the planned path may be relatively clumsy and lack optimization in terms of time and energy.

[0004] The Chinese patent application with publication number CN116698065A discloses a method and medium for path planning of an autonomous driving vehicle that integrates motion constraints and safety constraints. The method includes: 1) initializing path planning parameters; 2) establishing an evaluation function that introduces a safety distance and motion constraints; 3) using the evaluation function to calculate the shortest path of the autonomous driving vehicle; 4) screening the path nodes in the shortest path to remove redundant path nodes; 5) smoothing the path optimization nodes using a third-order Bezier curve to generate the planned path of the autonomous driving vehicle; 6) controlling the autonomous driving vehicle to drive according to the planned path. Each time the autonomous driving vehicle passes through a node, it is determined whether the obstacle has changed. If it has changed, the actual moving distance km value and the current position sstart of the vehicle are updated, and the current position is set as slast. The cost value of the affected nodes is updated, and then it returns to step 4). However, this method mainly considers obstacles and fails to comprehensively consider the overall situation of the road.

[0005] Therefore, the present invention proposes a method and system for generating complex-scene path planning based on multiple models. Summary of the Invention

[0006] The present invention aims to solve at least one of the technical problems existing in the prior art. For this purpose, the present invention proposes a method and system for generating complex-scene path planning based on multiple models, which can significantly improve the path planning ability of autonomous driving vehicles in complex scenes.

[0007] To achieve the above object, a method and system for generating complex-scene path planning based on multiple models are proposed. The method for generating complex-scene path planning based on multiple models includes the following steps:

[0008] Step 1: Pre-collect the static environment information set and dynamic environment information set in the road;

[0009] Step 2: Based on the static environment information set and dynamic environment information set, construct a multi-layer feature representation of the road;

[0010] Step 3: Based on the multi-layer feature representation of the road, construct a path planning model;

[0011] Step 4: Based on the path planning model, use an optimization model to generate a planned path;

[0012] The collection method of the static environment information set in the road is as follows:

[0013] Load the cloud map from the Internet and collect the obstacle point cloud data scanned by the lidar device;

[0014] Then, through the cloud map and the obstacle point cloud data, construct a road connectivity matrix;

[0015] The cloud map, the obstacle point cloud data, and the road connectivity matrix together form the static environment information set;

[0016] The method of constructing the road connectivity matrix through the cloud map and the obstacle point cloud data is as follows:

[0017] In the cloud map, extract the center line data of all lanes. Each center line consists of a series of discrete points;

[0018] Take each discrete point as the basic node of the road where the center line is located; these basic nodes will be used to describe the discretized structure of the road network;

[0019] In the cloud map, filter out the intersection positions and add convergence nodes at the intersections to represent the connection relationships between multiple roads;

[0020] The two-dimensional coordinates of the basic nodes and the convergence nodes form the road node set;

[0021] Mark any two adjacent nodes in the road node set as i and j respectively;

[0022] Mark the connectivity label value of the i-th node and the j-th node in the road connectivity matrix as Aij;

[0023] If the i-th node and the j-th node are continuously located on the same lane, set Aij = 1; if the i-th node and the j-th node are not continuously located on the same lane, set Aij = 0;

[0024] For each node i or node j in the set of road nodes, calculate the distance to the nearest obstacle. If the distance is less than the distance threshold of the obstacle, it is determined that the node is near the obstacle; otherwise, it is determined that there is no obstacle near the node.

[0025] For any pair of nodes i and j with Aij = 1, if any one of the nodes between node i and node j is near an obstacle, set Aij = 0.

[0026] The collection method of the dynamic environment information set is as follows:

[0027] Real-time point cloud data of dynamic obstacles in the road is collected in real-time through lidar and camera sensing devices, and the detected dynamic obstacles are tracked in real-time through a multi-object tracking algorithm to calculate their movement speed and direction.

[0028] Real-time traffic information of the road is obtained by introducing external real-time traffic data services.

[0029] The real-time point cloud data, movement speed and direction, and real-time traffic information of dynamic obstacles together constitute the dynamic environment information set.

[0030] The method of constructing the multi-layer feature representation of the road based on the static environment information set and the dynamic environment information set is as follows:

[0031] For any node i in the set of road nodes:

[0032] Use the degree centrality formula of the node to calculate the degree centrality of each node.

[0033] The degree centrality is the number of nodes directly connected to each node. The degree centrality of the node is used as a supplement to the road connectivity matrix to form the global topological feature.

[0034] Extract various geometric attributes of the road between every two directly connected nodes in the set of road nodes from the cloud map to form the local topological feature.

[0035] For each node in the set of road nodes:

[0036] Obtain the position and movement trajectory of each dynamic obstacle from the lidar and camera sensing devices.

[0037] For each dynamic obstacle, use the trajectory fitting method to predict its future trajectory and generate a probability distribution.

[0038] Then, based on the probability distribution, apply the Gaussian kernel function to construct the dynamic risk field of each dynamic obstacle.

[0039] The trajectory probability distribution of each dynamic obstacle corresponding to the node and the dynamic risk field constitute the dynamic environment characteristics;

[0040] The multi-layer feature representation includes global topological features, local topological features, and dynamic environment characteristics;

[0041] The method of using the trajectory fitting method to predict its future trajectory and generate a probability distribution is as follows:

[0042] Extract the two-dimensional coordinates of the dynamic obstacle at the most recent n moments in the past to form the historical operation trajectory; n is a preset duration parameter;

[0043] Obtain the running speed and running direction of the dynamic obstacle through the two-dimensional coordinates of every two adjacent moments of the dynamic obstacle;

[0044] Then, based on the running speed of the dynamic obstacle at different moments, obtain the running acceleration of the dynamic obstacle;

[0045] Based on the constant acceleration model, according to the running speed, running direction, and running acceleration of the dynamic obstacle, predict the two-dimensional coordinates at the next moment to form the future trajectory;

[0046] Introduce a Gaussian distribution to the future trajectory to obtain the probability distribution of the future trajectory to adapt to the uncertainty of the position of the dynamic obstacle;

[0047] The method of introducing the Gaussian distribution is as follows:

[0048] Express the Gaussian distribution as: ; where x and y represent the abscissa and ordinate of the two-dimensional coordinates respectively, and the mean is the position of the dynamic obstacle predicted based on the constant acceleration model, and the covariance matrix represents the predicted variance. The diagonal terms of the covariance matrix are calculated based on the volatility of the historical trajectory and the model error, so as to generate the corresponding probability distribution for each future moment ;

[0049] The method of constructing the dynamic risk field of each dynamic obstacle by applying the Gaussian kernel function based on the probability distribution is as follows:

[0050] Mark the risk field as R(x,y), representing the risk value of each point in the scene space of the road;

[0051] Define a set of uniform grid points in the scene space, and the resolution of the grid points is the preset resolution ;

[0052] Initialize the risk value of all grid points to 0, that is, set all R(x,y) to 0;

[0053] For each future trajectory distribution ; calculate the risk value of the grid points using a Gaussian kernel function;

[0054] The method for constructing the path planning model based on the multi-layer feature representation of the road is as follows:

[0055] Mark the node sequence included in the path to be planned as P; the node sequence P contains all the nodes passed by the planned path, mark the number of any one of the nodes as k, and mark the number of the node after this node as k + 1;

[0056] Based on the global topological feature, local topological feature, and dynamic environment feature, construct a planning objective function formed by weighted summation of distance, smoothness, risk, and time cost, and construct corresponding road connectivity constraints, dynamic risk constraints, and physical feasibility constraints to form a set of constraint conditions;

[0057] Taking the minimization of the planning objective function as the goal and the set of constraint conditions as the constraint set, a convex optimization problem formed is used as the path planning model;

[0058] The method for generating a planned path using the optimization model based on the path planning model is as follows:

[0059] By using an optimization algorithm to solve the path planning model, obtain the solved node sequence P as the planned path.

[0060] A complex scenario path planning generation system based on multiple models is proposed, including an environmental data collection module, a feature extraction module, a model construction module, and a path planning module; among them, each module is connected electrically;

[0061] The environmental data collection module pre-collects a set of static environmental information and a set of dynamic environmental information in the road, and sends the set of static environmental information and the set of dynamic environmental information to the feature extraction module;

[0062] The feature extraction module constructs a multi-layer feature representation of the road based on the set of static environmental information and the set of dynamic environmental information, and sends the multi-layer feature representation to the model construction module;

[0063] The model construction module constructs a path planning model based on the multi-layer feature representation of the road, and sends the path planning model to the path planning module;

[0064] The path planning module generates a planned path using the optimization model based on the path planning model.

[0065] Compared with the prior art, the beneficial effects of the present invention are:

[0066] The present invention first collects a set of static environment information and a set of dynamic environment information, then constructs a multi-layer feature representation of the road based on this information, including global topological features, local topological features, and dynamic environment features. Then, a path planning model is constructed based on the multi-layer feature representation. The objective function of this model comprehensively considers distance, smoothness, risk, and time cost, and the constraint conditions include road connectivity constraints, dynamic risk constraints, and physical feasibility constraints. Finally, an optimization algorithm is used to solve this model to generate the final planned path. By constructing a multi-layer feature representation, static and dynamic environment information is fused, and at the same time, the global topological structure, local geometric characteristics, and dynamic risks are considered, so that a safer, smoother, and more efficient path planning scheme can be generated, which can significantly improve the path planning ability of autonomous vehicles in complex scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 It is a flowchart of the method for generating a complex scenario path plan based on multiple models in Embodiment 1 of the present invention;

[0068] Figure 2 It is a diagram of the module connection relationship of the system for generating a complex scenario path plan based on multiple models in Embodiment 2 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0069] The technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative work fall within the scope of protection of the present invention.

[0070] Embodiment 1

[0071] As Figure 1 shown, the method for generating a complex scenario path plan based on multiple models includes the following steps:

[0072] Step 1: Collect a set of static environment information and a set of dynamic environment information in the road in advance;

[0073] Step 2: Construct a multi-layer feature representation of the road based on the set of static environment information and the set of dynamic environment information;

[0074] Step 3: Construct a path planning model based on the multi-layer feature representation of the road;

[0075] Step 4: Generate a planned path using an optimization model based on the path planning model;

[0076] In the implementation manner of the present application, the set of static environment information refers to the long-term fixed characteristics related to the road and the environment in the path planning scenario;

[0077] The dynamic environment information set refers to the information related to the dynamic changes in the path planning scenario.

[0078] Among them, the collection method of the static environment information set in the road is as follows:

[0079] In the implementation manner of this application, the main sources of the static environment information set are the cloud map loaded from the Internet and the obstacle point cloud data scanned by lidar devices;

[0080] Then, through the cloud map and the obstacle point cloud data, a road connectivity matrix is constructed;

[0081] The cloud map, the obstacle point cloud data, and the road connectivity matrix together constitute the static environment information set;

[0082] Specifically, the cloud map includes, but is not limited to, road topology information such as the connectivity between roads, intersection positions, lane distributions, etc., road geometric characteristics such as road curvature, slope, width, lane line positions, etc., fixed obstacle information including the positions of stationary obstacles (such as buildings, street lights, guardrails, isolation belts, etc.), and traffic rule information including road speed limits, one-way streets, no-go areas, signal light positions, etc.;

[0083] The obstacle point cloud data generated by lidar generally includes the two-dimensional or three-dimensional model of the obstacle, the position of the obstacle, etc.;

[0084] It should be understood that in current transportation vehicles (motor vehicles, aircraft, etc.), connecting to the Internet through a wireless network to download the latest cloud map is a mature technology, and installing some lidar in transportation vehicles, detecting obstacles through radar signals, and mapping the point cloud data of obstacles are also conventional means in this field. Therefore, the collection of the cloud map and the obstacle point cloud data is a mature existing technology, and this invention will not elaborate on it here;

[0085] By extracting the above static environmental information set, long-term reliable basic environmental data for path planning can be provided, including road topology information, geometric characteristics, fixed obstacle information, and traffic rule information. Specifically, the use of these features can significantly improve the stability and global optimization ability of path planning. Road topology information can describe the global connectivity of the road network, helping the path planning algorithm quickly generate a globally optimal path and avoiding the planning of invalid areas such as dead ends. Road geometric characteristics (such as curvature, slope, width) provide physical constraint conditions for path planning, ensuring that the generated path meets the dynamic constraints of the vehicle and avoiding the dangers brought by sharp turns and excessive slopes. In addition, fixed obstacle information can mark impassable areas, reducing the dependence on the dynamic perception module in path planning and improving the planning efficiency and accuracy. And traffic rule information (such as speed limits, one-way streets, etc.) can constrain the behavior of path planning to ensure that the path meets the requirements of traffic regulations. The extraction and comprehensive use of these features can enhance the globality, stability, and compliance of path planning, laying a solid foundation for path generation in complex scenarios;

[0086] In an embodiment of the present application, the road connectivity matrix is a mathematical representation form that describes the connectivity relationship between adjacent nodes in a road;

[0087] Thus, the method of constructing a road connectivity matrix through the cloud map and obstacle point cloud data is as follows:

[0088] In the cloud map, extract the center line data of all lanes. Each center line is composed of a series of discrete points. For example, a center line can be expressed as {(x1,y1),(x2,y2)...}, where (x1,y1) and (x2,y2) respectively represent two adjacent two-dimensional coordinate points on a center line, and the distance between every two adjacent two-dimensional coordinate points is determined according to actual needs;

[0089] Take each discrete point as the basic node of the road where the center line is located; these basic nodes will be used to describe the discretized structure of the road network;

[0090] In the cloud map, filter out the intersection positions and add convergence nodes at the intersections to represent the connection relationship between multiple roads;

[0091] The two-dimensional coordinates of the basic nodes and the convergence nodes form a road node set;

[0092] In an optional embodiment, for multi-lane roads, when the distance between adjacent lanes is small (judged by setting a lane distance threshold), node connections can also be added to represent lane changes. If the geometric center lines of two lanes are C1 and C2 respectively, then a lateral connection is inserted between the nodes of adjacent lanes;

[0093] Mark any two adjacent nodes in the set of road nodes as i and j respectively;

[0094] Mark the connectivity label value of the i-th node and the j-th node in the road connectivity matrix as Aij;

[0095] If the i-th node and the j-th node are continuously located on the same lane, set Aij = 1; if the i-th node and the j-th node are not continuously located on the same lane, set Aij = 0;

[0096] It should be understood that the setting process of this connectivity label value only describes the connectivity on the road center line and does not yet consider the influence of obstacles or cross lanes.

[0097] To this end, according to the type of obstacles, set a distance threshold for each type of obstacle; it should be noted that this distance threshold means that when the distance between the traffic vehicle and the obstacle is less than this distance threshold, it is considered that the traffic vehicle is difficult to pass, and thus it is judged that this node is impassable;

[0098] For each node i or node j in the set of road nodes, calculate the distance to the obstacle closest to it. If the distance is less than the distance threshold of this obstacle, judge that this node is near the obstacle; otherwise, judge that there is no obstacle near this node;

[0099] For any pair of nodes i and j with Aij = 1, if any one of the nodes between node i and node j is near an obstacle, set Aij = 0;

[0100] It should be noted that the generation of the above road connectivity matrix only considers one embodiment of obstacles. In the actual road traffic process, the influence of traffic rules also needs to be considered;

[0101] In a further feasible embodiment, the connectivity matrix can also be further adjusted according to known traffic rules (such as one-way roads, prohibited areas) in the cloud map. For example, if the connectivity between node i and node j violates traffic rules (such as opposite one-way traffic directions), set Aij = 0;

[0102] It can be understood that the road connectivity matrix can accurately reflect the connectivity between nodes and is dynamically adjusted in combination with obstacle information. When there is an obstacle in the middle of the road, the path planning algorithm will automatically bypass this area to ensure the safety of the planned path;

[0103] Furthermore, the collection method of the dynamic environment information set is as follows:

[0104] Real-time point cloud data of dynamic obstacles in the road is collected in real time by lidar and camera sensing devices, and the detected dynamic obstacles are tracked in real time by a multi-object tracking algorithm to calculate their moving speed and direction; specifically, the multi-object tracking algorithm is SORT or DeepSORT;

[0105] Real-time traffic information of the road is obtained by introducing external real-time traffic data services; specifically, the external real-time traffic data services such as traffic management platforms, navigation platform APIs, and the real-time traffic information includes but is not limited to real-time congestion conditions, construction areas, traffic accidents, etc.;

[0106] The real-time point cloud data, moving speed and direction, and real-time traffic information of dynamic obstacles together constitute a dynamic environment information set;

[0107] Considering that during the road driving process, the road feasibility is also affected by weather conditions, therefore, in a preferred embodiment, the dynamic environment information set may further include real-time weather data;

[0108] By extracting the dynamic environment information set in real time, it can provide the core information of the dynamic scene for the path planning system, including dynamic obstacle information, trajectory prediction information, road real-time conditions, and environmental impact information. The use of this information can significantly improve the real-time performance and safety of path planning in a dynamic environment. Dynamic obstacle information (such as the position information, speed, and direction of pedestrians and vehicles) can support the obstacle avoidance algorithm to adjust the path in real time and avoid collisions with obstacles. Trajectory prediction information helps the path planning system perceive potential risks in advance and achieve a safer path design by providing the future position distribution of dynamic obstacles. Road real-time conditions (such as congestion and accident areas) can provide a reference for the current optimal passing area for path planning, thereby improving the overall planning efficiency and passing time. And environmental impact information (such as weather and slippery roads) further enhances the safety assessment ability of path planning and supports the system to select a safer path under bad weather conditions. The extraction and real-time use of these dynamic environment features enable the path planning system to have excellent flexibility and adaptability in changing scenarios and provide important support for dynamic path optimization in complex scenarios.

[0109] In the implementation manner of the present application, the multi-layer feature representation is constructed through the following three levels:

[0110] Based on the static environment information set, the global topological features describing the overall connectivity and structural characteristics of the road;

[0111] Based on the static environment information set, the local topological features describing the geometric attributes of a single road segment, such as curvature, slope, width, etc.;

[0112] Describe the distribution of dynamic obstacles and the dynamic environmental characteristics of the risk area in combination with the dynamic environmental information set;

[0113] The characteristics of these three levels can provide information support from global to local and dynamically combine static and dynamic. By fusing these characteristics, a multi-level feature representation is formed.

[0114] Specifically, the method of constructing the multi-level feature representation of the road based on the static environmental information set and the dynamic environmental information set is as follows:

[0115] For any node i in the road node set:

[0116] Calculate the degree centrality of each node using the degree centrality formula of the node;

[0117] It should be noted that the degree centrality is the number of nodes directly connected to each node (i.e., Aij = 1). The degree centrality of the node is used as a supplement to the road connectivity matrix to form a global topological feature for evaluating the global influence of road nodes;

[0118] Extract various geometric attributes of the road between every two directly connected nodes in the road node set from the cloud map to form local topological features;

[0119] Specifically, the geometric attributes include but are not limited to curvature, slope, and road width. Since the cloud map Figure 1 is generally pre-surveyed, these geometric attributes can be directly extracted from the map;

[0120] For each node in the road node set:

[0121] Obtain the positions and movement trajectories of each dynamic obstacle from lidar and camera sensing devices;

[0122] For each dynamic obstacle, use the trajectory fitting method to predict its future trajectory and generate a probability distribution;

[0123] Then, based on the probability distribution, apply the Gaussian kernel function to construct the dynamic risk field of each dynamic obstacle;

[0124] The trajectory probability distribution and dynamic risk field of each dynamic obstacle corresponding to the node constitute the dynamic environmental characteristics;

[0125] The multi-level feature representation includes global topological features, local topological features, and dynamic environmental characteristics;

[0126] Specifically, the method of using the trajectory fitting method to predict its future trajectory and generate a probability distribution is as follows:

[0127] Extract the two-dimensional coordinates of the dynamic obstacle at the most recent n moments in the past to form a historical running trajectory; n is a preset duration parameter;

[0128] Obtain the running speed and running direction of the dynamic obstacle through the two-dimensional coordinates of every two adjacent moments of the dynamic obstacle;

[0129] Then, based on the running speed of the dynamic obstacle at different moments, obtain the running acceleration of the dynamic obstacle;

[0130] Based on the constant acceleration model, according to the running speed, running direction and running acceleration of the dynamic obstacle, predict the two-dimensional coordinates at the next moment to form a future trajectory;

[0131] Introduce a Gaussian distribution to the future trajectory to obtain the probability distribution of the future trajectory to adapt to the uncertainty of the position of the dynamic obstacle;

[0132] Specifically, the way of introducing the Gaussian distribution is as follows:

[0133] Represent the Gaussian distribution as: ; where x and y represent the abscissa and ordinate of the two-dimensional coordinates respectively, and the mean value is the position of the dynamic obstacle predicted based on the constant acceleration model, and the covariance matrix represents the predicted variance, which is used to quantify the prediction uncertainty. The diagonal terms of the covariance matrix can be calculated based on the volatility of the historical trajectory and the model error, so as to generate the corresponding probability distribution for each future moment ;

[0134] In the implementation manner of this application, a spatial risk field is generated according to the trajectory distribution of the obstacle, representing the dynamic risk in path planning. The higher the value of the risk field, the greater the risk of the area;

[0135] Further, the way of constructing the dynamic risk field of each dynamic obstacle by applying the Gaussian kernel function based on the probability distribution is as follows:

[0136] Mark the risk field as R(x, y), representing the risk value of each point in the scene space of the road;

[0137] Define a set of uniform grid points in the scene space, and the resolution of the grid points is a preset resolution ;

[0138] Initialize the risk values of all grid points to 0, that is, set all R(x, y) to 0;

[0139] For each future trajectory distribution ; calculate the risk value of the grid point using the Gaussian kernel function:

[0140] ; where \(w\) is the preset risk weight of the obstacle. The preset risk weight of the obstacle can be set for each type of obstacle in advance according to the type of the obstacle, and this risk weight represents the degree of risk that may be brought by touching this type of obstacle; \(u_x\) and \(u_y\) are the central point positions of the obstacle. The obstacle includes a stationary obstacle and a moving obstacle. For a moving obstacle, the central point position of the obstacle is the position predicted according to its running trajectory. Thus, the risk value of the grid point quantifies the risk contribution of the dynamic obstacle to the grid point, and combines the Gaussian distribution to describe the possibility and influence range of the predicted position of the obstacle;

[0141] Further, the method for constructing the path planning model based on the multi-layer feature representation of the road is as follows:

[0142] Mark the node sequence included in the path to be planned as \(P\); the node sequence \(P\) includes all the nodes passed by the planned path. Mark the number of any one of the nodes as \(k\), and the number of the node after this node as \(k + 1\);

[0143] It can be understood that by selecting different node sequences, the path planning algorithm can explore different path schemes. The selection of each node directly affects the total distance, smoothness, risk, and time cost of the path, the order and connection of the nodes, and the order of the selected node sequence determines the shape and direction of the path. Therefore, it is necessary to optimize the selection of the distance, smoothness, risk, etc. of the path;

[0144] Based on the global topological feature, local topological feature, and dynamic environment feature, construct a planning objective function formed by weighted summation of distance, smoothness, risk, and time cost, and construct corresponding road connectivity constraints, dynamic risk constraints, and physical feasibility constraints to form a set of constraint conditions;

[0145] Taking the minimization of the planning objective function as the goal and the set of constraint conditions as the constraint set, a convex optimization problem formed is used as the path planning model;

[0146] In the embodiment of the present application, the construction method of the planning objective function can be:

[0147] Mark the planning objective function as \(F_P(P)\);

[0148] Then the expression of the planning objective function \(F_P(P)\) is:

[0149] ;

[0150] where \(a_1\), \(a_2\), and \(a_3\) are all preset proportionality coefficients;

[0151] \(F_D(P)\) is the distance cost, and the distance cost ; where is the distance between the k-th node and the (k + 1)-th node; the distance cost represents the total distance that needs to be traveled for the planned path;

[0152] is the smoothness cost, and the smoothness cost ; where and are the curvature of the (k + 1)-th node and the curvature of the k-th node respectively; is the slope of the k-th node; Zk is the degree centrality of the k-th node, and b1, b2, and b3 are preset proportionality coefficients; this cost means that points with smaller curvature changes, smaller slopes, and closer to the center are preferably selected for movement;

[0153] is the risk cost, and the risk cost ; where is the risk value R(xk, yk) corresponding to the two-dimensional coordinates of the k-th node in the road, and xk and yk are the abscissa and ordinate of the k-th node respectively;

[0154] In a further preferred embodiment, the way of constructing the corresponding road connectivity constraint, dynamic risk constraint, and physical feasibility constraint to form a set of constraint conditions is as follows:

[0155] The road connectivity constraint is: ; is the connectivity label value between the k-th node and the (k + 1)-th node;

[0156] The dynamic risk constraint is ; where Rmax is a preset risk threshold;

[0157] The physical feasibility constraint is: ; Qmax is the maximum curvature value that the road can pass through preset;

[0158] Furthermore, the way of using the optimization model to generate a planned path based on the path planning model is as follows:

[0159] By using an optimization algorithm to solve the path planning model, the obtained node sequence P is obtained as the planned path;

[0160] Specifically, the optimization algorithm includes but is not limited to the ant colony algorithm, genetic algorithm, etc.

[0161] Embodiment 2

[0162] As Figure 2As shown, a multi-model based complex scenario path planning generation system includes an environmental data collection module, a feature extraction module, a model construction module, and a path planning module; among them, each module is electrically connected.

[0163] The environmental data collection module pre-collects a set of static environmental information and a set of dynamic environmental information in the road, and sends the set of static environmental information and the set of dynamic environmental information to the feature extraction module.

[0164] The feature extraction module constructs a multi-layer feature representation of the road based on the set of static environmental information and the set of dynamic environmental information, and sends the multi-layer feature representation to the model construction module.

[0165] The model construction module constructs a path planning model based on the multi-layer feature representation of the road, and sends the path planning model to the path planning module.

[0166] The path planning module generates a planned path using an optimization model based on the path planning model.

[0167] The specific implementation manner described above further details the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above is only the specific implementation manner of the present invention and is not used to limit the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

[0168] The above preset parameters or preset thresholds are set by those skilled in the art according to the actual situation or obtained through a large number of data simulations.

[0169] The above embodiments are only used to illustrate the technical method of the present invention and not to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical method of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A complex scene path planning generation method based on multiple models, characterized in that: The following steps are involved: Step 1: Pre-collect static environment information set and dynamic environment information set on the road; Step 2: Based on the static environment information set and the dynamic environment information set, construct a multi-layer feature representation of the road; Step 3: Construct a path planning model based on the multi-layer feature representation of the road; Step 4: Based on the path planning model, use the optimization model to generate a planned path; The method of constructing a multi-layer feature representation of a road based on a static environment information set and a dynamic environment information set is: For any node i in the road node set: Calculate the degree centrality of each node using the node's degree centrality formula; Degree centrality is the number of nodes directly connected to each node. The degree centrality of the node is used as a supplement to the road connectivity matrix to form the global topological feature; Extract various geometric attributes of the road between every two directly connected nodes in the road node set from the cloud map to form local topological features; For each node in the road node collection: Obtain the position and motion trajectory of each dynamic obstacle from the lidar and camera perception devices; For each dynamic obstacle, use the trajectory fitting method to predict its future trajectory and generate a probability distribution; Then, based on the probability distribution, the Gaussian kernel function is applied to construct the dynamic risk field of each dynamic obstacle; The trajectory probability distribution and dynamic risk field of each dynamic obstacle corresponding to the node constitute the dynamic environment characteristics; Multi-layer feature representation includes global topological features, local topological features and dynamic environment features.

2. The method for generating complex scene path planning based on multiple models according to claim 1, characterized in that: The static environment information set in the road is collected in the following manner: Load cloud maps from the Internet and collect point cloud data of obstacles scanned by the LiDAR device; Then, the road connectivity matrix is ​​constructed through cloud maps and obstacle point cloud data; Cloud maps, obstacle point cloud data, and road connectivity matrix together constitute a static environment information set.

3. The method for generating complex scene path planning based on multiple models according to claim 2, characterized in that: The method of constructing the road connectivity matrix through the cloud map and obstacle point cloud data is as follows: In the cloud map, the centerline data of all lanes are extracted, and each centerline consists of a series of discrete points; Each discrete point is regarded as the basic node of the road where the center line is located; these basic nodes will be used to describe the discretization structure of the road network; Filter out intersection locations in the cloud map and add convergence nodes at the intersections to indicate the connection relationship between multiple roads; The two-dimensional coordinates of the basic nodes and the convergence nodes constitute a road node set; Mark any two adjacent nodes in the road node set as i and j respectively; The connectivity label values ​​of the i-th node and the j-th node in the road connectivity matrix are marked as Aij; If the i-th node and the j-th node are continuously located on the same lane, set Aij=1; if the i-th node and the j-th node are not continuously located on the same lane, set Aij=0; For each node i or node j in the road node set, calculate the distance to the nearest obstacle. If the distance is less than the distance threshold of the obstacle, the node is judged to be near the obstacle. Otherwise, it is judged that there is no obstacle near the node. For any node pair consisting of node i and node j with Aij=1, if any node between node i or node j is located near an obstacle, set Aij=0.

4. The method for generating complex scene path planning based on multiple models according to claim 3, characterized in that: The dynamic environment information set is collected in the following manner: The laser radar and camera sensing devices are used to collect real-time point cloud data of dynamic obstacles on the road, and the detected dynamic obstacles are tracked in real time through the multi-target tracking algorithm to calculate their movement speed and direction; Obtain real-time traffic information of roads by introducing external real-time traffic data services; The real-time point cloud data of dynamic obstacles, movement speed and direction, and real-time traffic information together constitute a dynamic environment information set.

5. The method for generating complex scene path planning based on multiple models according to claim 4, characterized in that: The method of using the trajectory fitting method to predict its future trajectory and generate probability distribution is as follows: The two-dimensional coordinates of the dynamic obstacle at the latest n moments in the past are extracted to form the historical running trajectory; n is the preset time parameter; The running speed and running direction of the dynamic obstacle are obtained through the two-dimensional coordinates of the dynamic obstacle at every two adjacent moments; Then, based on the running speed of the dynamic obstacle at different times, the running acceleration of the dynamic obstacle is obtained; Based on the constant acceleration model, the two-dimensional coordinates of the next moment are predicted according to the running speed, running direction and running acceleration of the dynamic obstacle to form the future trajectory; Gaussian distribution is introduced to the future trajectory to obtain the probability distribution of the future trajectory to adapt to the uncertainty of the dynamic obstacle position.

6. The method for generating complex scene path planning based on multiple models according to claim 5, characterized in that: The method of introducing Gaussian distribution is: The Gaussian distribution is expressed as: ; where x and y represent the horizontal and vertical coordinates of the two-dimensional coordinates, respectively, and the mean is the position of the dynamic obstacle predicted based on the constant acceleration model, the covariance matrix Represents the variance of the forecast. The diagonal items of the covariance matrix are calculated based on the volatility of the historical trajectory and the model error, thereby generating the corresponding probability distribution for each future moment. .

7. The method for generating complex scene path planning based on multiple models according to claim 6, characterized in that: The method of constructing the dynamic risk field of each dynamic obstacle based on probability distribution and applying Gaussian kernel function is as follows: The risk field is marked as R(x,y), which represents the risk value of each point in the road scene space; Define a uniform set of grid points in scene space, with a resolution of the preset resolution ; Initialize the risk value of all grid points to 0, that is, set all R(x,y) to 0; For each future trajectory distribution ; Use Gaussian kernel function to calculate the risk value of grid points.

8. The method for generating complex scene path planning based on multiple models according to claim 7, characterized in that: The method of constructing the path planning model based on the multi-layer feature representation of the road is as follows: The node sequence included in the path to be planned is marked as P; the node sequence P includes all nodes passed by the planned path, and the number of any node is marked as k, and the number of the node after the node is marked as k+1; Based on global topological features, local topological features and dynamic environmental features, a planning objective function is constructed by weighted summation of distance, smoothness, risk and time cost, and corresponding road connectivity constraints, dynamic risk constraints and physical feasibility constraints are constructed to form a set of constraint conditions; A convex optimization problem consisting of minimizing the planning objective function is used as the path planning model.

9. The method for generating complex scene path planning based on multiple models according to claim 8, characterized in that: The method of generating a planned path based on the path planning model using the optimization model is as follows: By using an optimization algorithm to solve the path planning model, the solved node sequence P is obtained as the planned path.

10. A complex scene path planning generation system based on multiple models, which is used to implement the complex scene path planning generation method based on multiple models as described in any one of claims 1 to 9, characterized in that: It includes an environmental data collection module, a feature extraction module, a model building module and a path planning module; wherein each module is electrically connected; An environmental data collection module collects static environmental information sets and dynamic environmental information sets on the road in advance, and sends the static environmental information sets and dynamic environmental information sets to a feature extraction module; A feature extraction module constructs a multi-layer feature representation of the road based on a static environment information set and a dynamic environment information set, and sends the multi-layer feature representation to a model construction module; A model building module builds a path planning model based on the multi-layer feature representation of the road, and sends the path planning model to the path planning module; The path planning module generates a planned path based on the path planning model using the optimization model.

Images