Motion planning method and device for mobile robot in indoor dynamic scene

Abstract

The invention provides a mobile robot motion planning method and device for an indoor dynamic scene, and the method comprises the steps: generating an indoor environment map and the positioning information of a mobile robot according to environment perception data obtained in real time, and carrying out the denoising and classification of the environment perception data, static obstacle points and dynamic obstacle points are extracted according to the recognition features; constructing a comprehensive environment model of static obstacles and dynamic obstacles; identifying and marking the current position and the predicted position of the dynamic obstacle by utilizing a machine learning model based on the track prediction information of the dynamic obstacle in the comprehensive environment model, and further identifying potential navigation path key points of the mobile robot; constructing a multi-criterion preference function based on the navigation path key points, and measuring and serializing the navigation path key points to obtain local optimal navigation path key points; and determining a planned path by applying a branch and bound method.

Description

Technical Field

[0001] The present application relates to the technical field of mobile robot motion planning, and in particular to a method and device for mobile robot motion planning in indoor dynamic scenes. Background Art

[0002] With the rapid development of smart health care, smart home and service robots, mobile robots are playing an increasingly important role in home health care and social services. However, indoor environments usually have complex dynamic characteristics, such as moving obstacles (such as people, pets, furniture, etc.), narrow passages and changing environment layouts.

[0003] Existing mobile robot navigation methods mainly rely on pre-built environmental maps or static environment mapping and localization (SLAM) during navigation. However, these methods often have problems such as slow response speed, inflexible path planning, easy to fall into local optimality, and lack of dynamic obstacle avoidance when facing dynamic obstacles and real-time environmental changes. Summary of the invention

[0004] The embodiments of the present application provide a method and device for mobile robot motion planning in indoor dynamic scenes to improve the autonomy and adaptability of the mobile robot.

[0005] In a first aspect, an embodiment of the present application provides a method for motion planning of a mobile robot in an indoor dynamic scene, comprising:

[0006] Generate an indoor environment map and positioning information of the mobile robot based on the environmental perception data acquired in real time, denoise and classify the environmental perception data, and extract static obstacle points and dynamic obstacle points based on recognition features;

[0007] Construct a comprehensive environmental model of static and dynamic obstacles;

[0008] Based on the trajectory prediction information of the dynamic obstacle in the comprehensive environment model, a machine learning model is used to identify and mark the current position and predicted position of the dynamic obstacle, thereby identifying the potential key points of the navigation path of the mobile robot;

[0009] Based on the navigation path key points, a multi-criteria preference function is constructed, and each of the navigation path key points is measured and serialized to obtain a local optimal navigation path key point;

[0010] Apply the branch and bound method to determine the planning path;

[0011] Based on the local optimal navigation path key points and travel risk assessment indicators within the planned path range, a guided sampling method is used to generate a planned path to a target location;

[0012] Converting the planned path into an executable motion trajectory, and finally determining the navigation trajectory of the mobile robot;

[0013] During the movement of the mobile robot, the planned path and motion trajectory are dynamically adjusted according to the real-time changing environmental perception data.

[0014] In a feasible implementation, the building of a comprehensive environment model of static obstacles and dynamic obstacles includes:

[0015] Building an occupancy grid map of the static environment based on the static obstacle points, marking all inaccessible areas, and forming a static obstacle environment model;

[0016] Using a motion prediction module to predict the future position and motion trajectory of the dynamic obstacle point to form a dynamic obstacle prediction model;

[0017] The static obstacle environment model and the dynamic obstacle prediction model are integrated to generate the comprehensive environment model.

[0018] In a feasible implementation, the constructing of a multi-criteria preference function includes:

[0019] The weighted average method is used to integrate feature importance, movement cost and topological structure to construct a preference function;

[0020] The preference function is used to calculate the preference value of each navigation path key point, and the navigation path key point with the largest value is configured as the local optimal navigation path key point.

[0021] In a feasible implementation, the preference function is:

[0022] R(b i )=w ig IG i -w c ·C i +w topo ·T i (1);

[0023] Among them, R(b i ) represents the preference function, w ig ,w c ,w topo Respectively represent the weight coefficients of feature importance, movement cost and topological structure; IG i Indicates feature importance; C i represents the moving cost; T i Represents the topological structure.

[0024] In a feasible implementation, the method for calculating the feature importance includes:

[0025]

[0026] Among them, A new (b i ) indicates that the mobile robot moves from the current position to the key point b of the navigation path i After that, the newly explored environment area is A total Indicates that the mobile robot moves to the key point b of the navigation path i The total area that has been explored before, ε is a constant.

[0027] In a feasible implementation, the method for calculating the movement cost includes:

[0028] Calculate the mobile robot from the current position p robot Move to candidate boundary point b i The path cost includes:

[0029] C i = dist(p robot ,b i )(3);

[0030] Among them, dist(·,·) is the distance function, p robot Indicates the current position, b i It is the key point of the navigation path.

[0031] In a feasible implementation, the method of generating a planned path to a target location by using a guided sampling method based on the local optimal navigation path key points and the travel risk assessment index within the planned path range includes:

[0032] The travel risk assessment index is established on the geometric and semantic features of the local point cloud or grid map of the environment to evaluate whether a certain position p is safe for the mobile robot to pass;

[0033] The travel risk assessment index calculation function is:

[0034]

[0035] Among them, G(p) represents the travel risk assessment index, Δz(p) represents the height change within the neighborhood centered on p, to determine whether the ground is flat; h th is the maximum acceptable height difference threshold; α slope (p) is the average slope or maximum slope within the neighborhood, θ th is the passable slope threshold;

[0036] And / or, a comprehensive determination is made on whether the mobile robot can pass safely by combining normal vector distribution or semantic segmentation results.

[0037] In a feasible implementation, the method of generating a planned path to a target location by using a guided sampling method based on the local optimal navigation path key points and the travel risk assessment index within the planned path range includes:

[0038] The key points of the local optimal navigation path and the travel risk assessment index within the planned path range are combined to obtain a guidance set Ψ={(p, G(p)), ...} to provide feasibility guidance for subsequent path planning;

[0039] When adopting the path planning method based on guided sampling, the points falling in the guided set Ψ are sampled preferentially or forcibly in the sampling space to reduce the exploration of inaccessible or low-access areas, enhance the sampling density of the space near the target key points, and improve the real-time and adaptability of the path.

[0040] In a feasible implementation, converting the planned path into an executable motion trajectory and finally determining the navigation trajectory of the mobile robot includes:

[0041] The generated planned path is discretized into several path points, the direction and posture of each path point are calculated, and the navigation trajectory is generated by combining the kinematic and dynamic models of the mobile robot.

[0042] Get the initial discrete trajectory Finally, the multi-constraint penalty optimization method J comprehensively considers the safety of environmental obstacles, the smoothness of the trajectory and the minimum energy consumption of the robot to construct the objective function J(τ):

[0043]

[0044] Where T is the execution time of the entire trajectory; f obs (x, y) is used to measure the distance cost between the current position (x, y) and the obstacle; κ(t) represents the trajectory curvature, which is represented by Calculate, x′, x″, y′, y″ are the first and second order derivatives of the trajectory with respect to time; a(t) represents the linear acceleration or angular acceleration of the robot, which is expressed in discrete form by |v i+1 -v i | / Δt or |ω i+1 -ω i | / Δt metric; w obs ,w curv ,w acc is the weight coefficient of each penalty;

[0045] The navigation trajectory is iteratively updated for multiple rounds using gradient descent, heuristic search, or sampling-based optimization methods to gradually reduce the value of J(τ) while satisfying the constraints of velocity, acceleration, and obstacle collision to obtain the final motion planning trajectory.

[0046] In a second aspect, an embodiment of the present application provides a mobile robot motion planning device for an indoor dynamic scene, applying the mobile robot motion planning method for an indoor dynamic scene described in the first aspect, including:

[0047] Controller;

[0048] A data processing module, electrically connected to the controller, the data processing module is used to generate an indoor environment map and real-time location information of the mobile robot according to the environmental perception data acquired in real time, denoise and classify the environmental perception data, and extract static obstacle points and dynamic obstacle points according to recognition features;

[0049] A model building module is electrically connected to the controller, and the model building module builds an occupancy grid map of the static environment based on the static obstacle points, marks all inaccessible areas, builds a static obstacle environment model, predicts the future position and motion trajectory of the dynamic obstacle points based on the dynamic obstacle points, and builds a dynamic obstacle prediction model; finally, the static obstacle environment model is integrated with the dynamic obstacle prediction model to generate a comprehensive environment model;

[0050] A function construction and calculation model, electrically connected to the controller, the function construction and calculation model is used to construct a multi-criteria preference function based on the navigation path key points, and measure and serialize each of the navigation path key points to obtain a local optimal navigation path key point;

[0051] A path planning module is electrically connected to the controller, and the path planning module is used to determine the planned path by applying a branch and bound method; and based on the local optimal navigation path key points and the travel risk assessment index within the planned path range, a guided sampling method is used to generate a planned path to a target position; and the path planning module can dynamically adjust the planned path of the mobile robot according to the real-time changing environment and the dynamic obstacle position;

[0052] A path conversion and optimization module is electrically connected to the controller, and is used to convert the planned path into an executable motion trajectory, optimize the navigation trajectory through a multi-constraint optimization algorithm, and finally generate the motion planning path of the mobile robot in combination with environmental safety, trajectory smoothness and energy consumption minimization.

[0053] In a first aspect, an embodiment of the present application provides a method for motion planning of a mobile robot in an indoor dynamic scene, comprising:

[0054] Generate indoor environment maps and mobile robot positioning information based on real-time environmental perception data, denoise and classify environmental perception data, extract static obstacle points and dynamic obstacle points based on identification features, and build a comprehensive environmental model of static obstacles and dynamic obstacles. By converting complex environmental information into an easy-to-process comprehensive environmental model including static obstacles and dynamic obstacles, the computing and storage costs of mobile robots are significantly reduced, and the computing power and task processing speed of mobile robots are improved.

[0055] Based on the trajectory prediction information of dynamic obstacles in the comprehensive environment model, the machine learning model is used to identify and mark the current position and predicted position of the dynamic obstacles, thereby identifying the potential key points of the navigation path of the mobile robot. Since the machine learning model can be used to identify and mark the current position and predicted position of the dynamic obstacles, the position of the dynamic obstacles can be identified and dynamically updated, thereby improving the mobile robot's ability to adapt to and respond to dynamic environments.

[0056] Based on the key points of the navigation path, a multi-criteria preference function is constructed, and each key point of the navigation path is measured and serialized to obtain the local optimal key point of the navigation path. By using the preference function to measure and serialize each key point of the navigation path, the local optimal key point of the navigation path is selected to ensure the rationality of the selection of the key point of the navigation path, improve the global optimality of the path planning, and avoid short-sighted and greedy exploration.

[0057] The branch and bound method is applied to determine the planning path, and an efficient path covering all key points of the unexplored path is planned to achieve comprehensive exploration and coverage of unknown areas by the mobile robot.

[0058] Based on the local optimal navigation path key points and travel risk assessment indicators within the planned path range, a guided sampling method is used to generate a planned path to the target location; the planned path is converted into an executable motion trajectory, and the navigation trajectory of the mobile robot is finally determined; during the movement of the mobile robot, the planned path and motion trajectory are dynamically adjusted according to the real-time changing environmental perception data. Real-time monitoring of environmental changes and dynamic obstacle positions, combined with dynamic replanning and trajectory adjustment mechanisms, ensures the robot's safe and stable navigation in complex dynamic environments. At the same time, the machine learning model algorithm is used to predict and optimize environmental changes, improve the response speed and accuracy of path planning and trajectory optimization, and enhance the adaptability of mobile robots in changing environments.

[0059] In the second aspect, the embodiment of the present application provides a mobile robot motion planning device for indoor dynamic scenes, which applies the mobile robot motion planning method for indoor dynamic scenes described in the first aspect, and specifically includes a controller, a data processing module, a model building module, a function building and calculation model, a path planning module, and a path conversion and optimization module. Among them, the data processing module is used to generate an indoor environment map and real-time location information of the mobile robot based on the real-time acquired environmental perception data, denoise and classify the environmental perception data, and extract static obstacle points and dynamic obstacle points based on the recognition features. The model building module constructs an occupancy grid map of the static environment based on static obstacle points, marks all inaccessible areas, builds a static obstacle environment model, predicts the future position and motion trajectory of dynamic obstacle points, and builds a dynamic obstacle prediction model; finally, the static obstacle environment model and the dynamic obstacle prediction model are integrated to generate a comprehensive environment model; the function construction and calculation model is used to construct a multi-criteria preference function based on the key points of the navigation path, and measure and serialize each navigation path key point to obtain the local optimal navigation path key point; the path planning module uses the branch and bound method to determine the planned path; and based on the local optimal navigation path key points and travel risk assessment indicators within the planned path range, the guided sampling method is used to generate a planned path to the target position; and the path planning module can dynamically adjust the planned path of the mobile robot according to the real-time changing environment and dynamic obstacle positions.

[0060] The path conversion and optimization module is used to convert the planned path into an executable motion trajectory. Through a multi-constraint optimization algorithm, it combines environmental safety, trajectory smoothness and energy consumption minimization to optimize the navigation trajectory and finally generate the mobile robot motion planning path. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] The drawings described herein are used to provide further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present application and do not constitute improper limitations on the present invention.

[0062] In the attached picture:

[0063] Figure 1 is a schematic diagram of a mobile robot motion planning method for an indoor dynamic scene provided by an embodiment of the present application;

[0064] Figure 2 is a schematic diagram of a mobile robot motion planning process provided by an embodiment of the present application;

[0065] Figure 3 It is a schematic diagram of a mobile robot motion planning device for indoor dynamic scenes provided by an embodiment of the present application.

[0066] Description of reference numerals:

[0067] 100-controller; 200-data processing module; 300-model building module; 400-function building and calculation model; 500-path planning module; 600-path conversion and optimization module. DETAILED DESCRIPTION

[0068] In order to enable those skilled in the art to better understand the technical solutions in the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. 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 the field without creative work should fall within the scope of protection of the present application.

[0069] In the description of the embodiments of the present application, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present application, "plurality" means at least two, for example, two, three, etc., unless otherwise clearly and specifically defined.

[0070] In this application, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements, unless otherwise clearly defined. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0071] In the present application, unless otherwise clearly specified and limited, a first feature being "above" or "below" a second feature may mean that the first and second features are in direct contact, or the first and second features are in indirect contact through an intermediate medium. Moreover, a first feature being "above", "above" or "above" a second feature may mean that the first feature is directly above or obliquely above the second feature, or simply means that the first feature is higher in level than the second feature. A first feature being "below", "below" or "below" a second feature may mean that the first feature is directly below or obliquely below the second feature, or simply means that the first feature is lower in level than the second feature.

[0072] With the rapid development of smart health care, smart home and service robots, mobile robots are playing an increasingly important role in home health care and social services. However, indoor environments usually have complex dynamic characteristics, such as moving obstacles (such as people, pets, furniture, etc.), narrow passages and changing environment layouts.

[0073] Existing mobile robot navigation methods mainly rely on pre-built environmental maps or static environment mapping and localization (SLAM) during navigation. However, these methods often have problems such as slow response speed, inflexible path planning, easy to fall into local optimality, and lack of dynamic obstacle avoidance when facing dynamic obstacles and real-time environmental changes.

[0074] In order to solve the above problems, an embodiment of the present application provides a mobile robot motion planning method for indoor dynamic scenes. The solution provided by the embodiment of the present application will be described in detail below in conjunction with the drawings in the specification.

[0075] Figure 1 It is a structural diagram of a method for motion planning of a mobile robot in an indoor dynamic scene provided by an embodiment of the present application; Figure 2 It is a schematic diagram of the motion planning process of a mobile robot provided in one embodiment of the present application.

[0076] Reference Figure 1 and Figure 2 As shown, in a first aspect, an embodiment of the present application provides a mobile robot motion planning method for an indoor dynamic scene, comprising:

[0077] S100: Generate an indoor environment map and positioning information of the mobile robot based on the environmental perception data acquired in real time, denoise and classify the environmental perception data, and extract static obstacle points and dynamic obstacle points based on recognition features.

[0078] The mobile robot is equipped with environmental feature acquisition components such as 2D laser radar, visual camera, depth measurement module, inertial measurement unit (IMU) and position sensor to obtain environmental perception data, such as the shape, size and color of objects in the environment. The mobile robot uses multi-sensor fusion technology to obtain real-time 3D point cloud data, image information and its own motion status data.

[0079] At the same time, the mobile robot exemplarily generates and updates the indoor environment map in real time by fusing multi-source sensor data through the SLAM algorithm, and accurately estimates the positioning information and posture of the mobile robot. In addition, the environmental perception data is denoised and classified, and static obstacle points and dynamic obstacle points are extracted according to the recognition features. Exemplarily, the recognition features may include semantic features, geometric features, normal features, posture features, and position features.

[0080] S200: Construct a comprehensive environment model of static obstacles and dynamic obstacles.

[0081] Exemplarily, it specifically includes constructing an occupancy grid map of the static environment based on the static obstacle points, marking all inaccessible areas, and forming a static obstacle environment model.

[0082] The motion prediction module is used to predict the future position and motion trajectory of dynamic obstacles to form a dynamic obstacle prediction model. The static obstacle environment model is fused with the dynamic obstacle prediction model to generate the comprehensive environment model. Exemplarily, a Kalman filter or a particle filter can be used to predict the future position and motion trajectory of dynamic obstacle points. It should be noted that the technology of fusing the static obstacle environment model with the dynamic obstacle prediction model is a prior art and will not be described in detail here. The motion prediction module is used to predict the trajectory of dynamic obstacles, which can accurately identify and dynamically update the position of dynamic obstacles, thereby improving the responsiveness of the mobile robot to the dynamic environment.

[0083] S300: Based on the trajectory prediction information of the dynamic obstacle in the comprehensive environment model, a machine learning model is used to identify and mark the current position and predicted position of the dynamic obstacle, thereby identifying potential key points of the navigation path of the mobile robot.

[0084] By analyzing the dynamic obstacle trajectory prediction information in the comprehensive environment model, the new position and movement trend of the dynamic obstacles are identified and updated in real time through the machine learning model, thereby identifying the potential key points of the navigation path of the mobile robot. In addition, based on the distribution variance information of dynamic obstacles and static obstacles, the potential free space area is identified, and combined with the indoor environment map, the key point set of the explored and unexplored areas is determined.

[0085] S400: Based on the navigation path key points, a multi-criteria preference function is constructed, and each navigation path key point is measured and serialized to obtain a local optimal navigation path key point.

[0086] Exemplarily, a weighted average method is used to integrate feature importance, movement cost, and topological structure to construct a preference function. Feature importance is used to represent the area of ​​the new environment that can be explored through this point. Movement cost is used to represent the distance from the current position of the mobile robot to the target point, and topological structure represents path connectivity, which can evaluate the degree of association between the target key point and the previously selected key point, so as to avoid the navigation path key point selection process falling into the local optimum.

[0087] The preference function based on multiple criteria can ensure the rationality of the selection of navigation key points, improve the global optimality of path planning, and avoid short-sighted and greedy exploration.

[0088] Specifically, the preference function is:

[0089] R(b i )=w ig IG i -w c ·C i +w topo ·T i (1);

[0090] Among them, R(b i ) represents the key point b of the navigation path i The preference function, w ig ,w c ,w topo Respectively represent the weight coefficients of feature importance, movement cost and topological structure; IG i Indicates feature importance; C i represents the moving cost; T i Represents the topological structure.

[0091] The calculation methods of feature importance include:

[0092]

[0093] Among them, A new (b i ) indicates that the mobile robot moves from the current position to the key point b of the navigation path i After that, the newly explored environment area is A total Indicates that the mobile robot moves to the key point b of the navigation path i The total area that has been explored before, ε is a constant.

[0094] The calculation method of moving cost includes:

[0095] Calculate the mobile robot from the current position p robot Move to candidate boundary point b i The path cost includes:

[0096] C i = dist(p robot ,b i )(3);

[0097] Among them, dist(·,·) is the distance function, p robot Indicates the current position, b i It is the key point of the navigation path.

[0098] Topology i The calculation method is:

[0099]

[0100] Among them, Ω represents the set of boundary points that have been selected or visited, conn(b i ,b j ) Characterization b i With b j Whether they are in the same connected area (which can be defined as 1 for connected, 0 for disconnected, or a score between 0 and 1 according to the length of the connected path), |Ω| represents the size of the set. i The larger the value, the greater the b i The better the connectivity with the explored areas, the more it helps maintain the consistency and traceability of the global path.

[0101] After the preference value calculation is completed, all navigation path key points {b1, b2, ..., b n According to R(b i ) values ​​are sorted from large to small, and several navigation path key points with the highest preference values ​​are selected for priority access or further path planning. If you need to select only one local optimal navigation path key point from multiple boundary points, you can directly select R(b i )The largest one is used as the current navigation target, configured as the key point of the local optimal navigation path, and used as the current navigation target of the mobile robot.

[0102] S500: Applying a branch and bound method to determine a planning path.

[0103] The branch and bound method is applied to plan an efficient path covering all unexplored boundary points, ensuring that the robot can fully cover the unknown area. It should be noted that the branch and bound method is a prior art and a conventional technique in the field, and will not be described in detail here.

[0104] S600: Based on the local optimal navigation path key points and the travel risk assessment index within the planned path range, a guided sampling method is used to generate a planned path to the target location.

[0105] Exemplarily, based on the key points of the local optimal navigation path within the planned path and the travel risk assessment index, a guidance set including local exploration targets and inaccessible areas is established to provide guidance information for path planning. In this process, it is necessary to quantitatively evaluate the travel risk assessment index. The present application establishes the travel risk assessment index G(p) based on the geometric and semantic features of the local point cloud or grid map of the environment to evaluate whether a certain position p is safe for the robot to pass.

[0106] The calculation function of the travel risk assessment index is:

[0107]

[0108] Wherein, Δz(p) represents the height change within a certain neighborhood (e.g., radius r) centered at p, such as the maximum height difference or standard deviation, which is used to determine whether the ground is flat; h th is the maximum acceptable height difference threshold; α slope (p) is the average slope or maximum slope in the neighborhood, θ th is the passable slope threshold (such as 10 degrees or 15 degrees); when the environment contains carpets, gaps or other details, it can be further combined with the normal vector distribution or semantic segmentation results for comprehensive judgment.

[0109] Based on the binary passability judgment defined by G(p) in the above formula (5), it can also be expanded to a continuous numerical passability score, such as G(p)∈[0,1], to characterize the passability at different locations. Then, the travel risk assessment index is combined with the information of the key points of the local optimal navigation path within the planned path range to obtain the guidance set Ψ={(p,G(p)),…}, which provides passability guidance for subsequent path planning.

[0110] In addition, when using a path planning method based on guided sampling (such as RRT*), the points that fall in the Ψ set can be sampled preferentially or forcibly in the sampling space, thereby reducing the exploration of impassable or low-passage areas and increasing the sampling density of the space near the target key points, thereby improving the real-time and adaptability of the path. The specific process is as follows:

[0111] 1) When constructing RRT* or other sampling trees, first select several key points with high accessibility scores G(p) from Ψ as seed nodes;

[0112] 2) During random sampling or iterative expansion, if the new expansion node can be connected to these seed nodes, the expansion branch will be retained first;

[0113] 3) Repeat the iteration until a path is found from the current position p of the mobile robot. robot To the global optimal key point b * The feasible path or the iteration limit is reached;

[0114] 4) Output the final planned path for subsequent trajectory generation and optimization.

[0115] S700: Convert the planned path into an executable motion trajectory, and finally determine the navigation trajectory of the mobile robot.

[0116] The generated planned path is discretized into several path points, the direction and posture of each path point are calculated, and the initial executable trajectory is generated by combining the robot's kinematic and dynamic models, including parameters such as speed, acceleration, and steering angle.

[0117] The generated planning path is discretized into several path points {P1, P2, ..., P m}, and calculate the direction and attitude of adjacent path points. Let P i =(x i ,y i ), then the direction angle θ of the mobile robot in the 2D plane i It can be obtained by the following formula:

[0118] θ i =arctan2(y i+1 -y i ,x i+1 -x i )(6);

[0119] When the mobile robot needs to consider the posture (Roll, Pitch, Yaw) of the mobile robot in a 3D environment, the complete posture parameters (φ i ,ψ i ,θ i ).

[0120] Then, the kinematic and dynamic models of the mobile robot (such as differential drive or omnidirectional drive) are combined to assign a suitable speed v to each path point. i , acceleration a i and the steering angular velocity ω i . etc., to satisfy

[0121]

[0122] Where Δt represents the discrete time step. The above formula can be adjusted appropriately according to the actual robot kinematic model. For example, the omnidirectional wheel model can also consider the lateral motion component.

[0123] In some examples, a trajectory optimization method with multiple constraints and penalties can be used to optimize the initial executable trajectory and generate a final navigation trajectory by comprehensively considering the safety of environmental obstacles, the smoothness of the trajectory, and the requirement to minimize the energy consumption of the robot.

[0124] Specifically, after obtaining the initial executable trace Finally, it is necessary to comprehensively consider the safety of environmental obstacles, the smoothness of the trajectory, and the minimum energy consumption of the robot through the optimization method of multi-constraint penalties. The following objective function can be constructed:

[0125]

[0126] Where T is the execution time of the entire trajectory; f obs(x, y) is used to measure the distance cost between the current position (x, y) and the obstacle; κ(t) represents the trajectory curvature, which is represented by Calculate, x′, x″, y′, y″ are the first and second order derivatives of the trajectory with respect to time; a(t) represents the linear acceleration or angular acceleration of the robot, which is expressed in discrete form by |v i+1 -v i | / Δt or |ω i+1 -ω i | / Δt metric; w obs ,w curv ,w acc is the weight coefficient of each penalty;

[0127] Finally, the trajectory is iteratively updated for multiple rounds using gradient descent, heuristic search or sampling-based optimization methods (such as Covariance Matrix Adaptation, CMA-ES), gradually reducing the value of J(τ) while satisfying constraints such as speed, acceleration and obstacle collision, thereby obtaining a smooth, safe and energy-efficient final navigation trajectory.

[0128] According to the optimized navigation trajectory, the mobile robot is controlled to move along the planned path to complete autonomous navigation in the environment. At this stage, the speed, acceleration and steering angle can be tracked and adjusted in real time based on sensor feedback, and the trajectory of the mobile robot can be dynamically planned to cope with sudden obstacles or environmental changes.

[0129] S800: During the movement of the mobile robot, the planned path and motion trajectory are dynamically adjusted according to the real-time changing environmental perception data.

[0130] Specifically, during the movement of the mobile robot, it obtains the latest environmental perception data in real time through sensors, dynamically updates the ring occupancy model and ground grid map, and then dynamically adjusts the planned path and motion trajectory.

[0131] The ring occupancy model refers to a local map model that focuses on updating the occupancy status within a certain radius (or several concentric rings) around the mobile robot, that is, with the current position of the mobile robot as the center, the radius r1, r2, ..., r k As the boundary, several concentric ring regions are formed. This model is denoted as O ring , which can be updated by using the occupancy probability calculation based on Bayesian filtering:

[0132]

[0133] Among them, m i Indicates the occupancy state of a grid (or spatial grid unit) (occupied or free), z tRepresents the measurement information of the sensor at time t. Since only the grids within the annular range are frequently updated, the frequency of large-scale updates to the entire global map can be reduced, while ensuring the real-time update of the local environment and reducing the robot's computing burden.

[0134] During the movement of the mobile robot, if it detects that there are new obstacles on the planned path or the movement trend of dynamic obstacles may affect the current path, the path will be replanned and the trajectory will be adjusted immediately. Specifically, it includes:

[0135] 1) Local conflict detection: Detect the shortest distance or time-to-collision with dynamic obstacles on the current optimal path. If it is lower than the set threshold, replanning is triggered.

[0136] 2) Path correction or re-solving: Based on the updated ring occupancy model O ring and global map, to quickly correct or heuristically search local paths (such as D*Lite);

[0137] 3) Trajectory reoptimization: Based on the new local path, the discrete trajectory is reoptimized to ensure that constraints such as speed, acceleration, and obstacle safety distance are met;

[0138] 4) Restore global planning: If the local path correction cannot meet the navigation requirements, it will fall back to the global planning layer and search for a new feasible global optimal path and trajectory under the comprehensive environment model;

[0139] 5) Execute updated trajectory: Output updated trajectory instructions to the robot's underlying controller to continue the navigation task.

[0140] Through the above steps, while ensuring the response speed of local environmental changes, the optimality and consistency of the global path can also be taken into account, ensuring that the robot can complete autonomous navigation tasks safely and efficiently in complex dynamic environments.

[0141] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0142] Based on the same inventive concept, the embodiment of the present application also provides a mobile robot motion planning device for indoor dynamic scenes for implementing the mobile robot motion planning method for indoor dynamic scenes involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in the embodiments of one or more mobile robot motion planning devices for indoor dynamic scenes provided below can refer to the limitations of the mobile robot motion planning method for indoor dynamic scenes above, and will not be repeated here.

[0143] Figure 3 It is a schematic diagram of a mobile robot motion planning device for indoor dynamic scenes provided by an embodiment of the present application.

[0144] Reference Figure 3 As shown, in the second aspect, the embodiment of the present application provides a mobile robot motion planning device for indoor dynamic scenes, including a controller 100, a data processing module 200, a model building module 300, a function building and calculation model 400, a path planning module 500 and a path conversion and optimization module 600. Among them, the data processing module 200 is used to generate an indoor environment map and real-time position information of the mobile robot according to the real-time acquired environmental perception data, denoise and classify the environmental perception data, and extract static obstacle points and dynamic obstacle points according to the recognition features. The model construction module 300 constructs an occupancy grid map of the static environment based on static obstacle points, marks all inaccessible areas, constructs a static obstacle environment model, predicts the future position and motion trajectory of dynamic obstacle points, and constructs a dynamic obstacle prediction model; finally, the static obstacle environment model and the dynamic obstacle prediction model are integrated to generate a comprehensive environment model; the function construction and calculation model 400 is used to construct a multi-criteria preference function based on the key points of the navigation path, and measure and serialize each key point of the navigation path to obtain the local optimal navigation path key point; the path planning module 500 uses the branch and bound method to determine the planned path; and based on the local optimal navigation path key points and the travel risk assessment index within the planned path range, a guided sampling method is used to generate a planned path to the target position; and the path planning module 500 can dynamically adjust the planned path of the mobile robot according to the real-time changing environment and the dynamic obstacle position.

[0145] The path conversion and optimization module 600 is used to convert the planned path into an executable motion trajectory, optimize the navigation trajectory through a multi-constraint optimization algorithm, and finally generate the mobile robot motion planning path by combining environmental safety, trajectory smoothness and energy minimization.

[0146] Each module in the above-mentioned mobile robot motion planning device for indoor dynamic scenes can be implemented in whole or in part by software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0147] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0148] It is easy to understand that those skilled in the art can combine, split, reorganize, etc. the embodiments of the present application to obtain other embodiments based on the several embodiments provided in the present application, and these embodiments do not exceed the protection scope of the present application.

[0149] The above specific implementation methods further explain in detail the purpose, technical solutions and beneficial effects of the embodiments of the present application. It should be understood that the above are only specific implementation methods of the embodiments of the present application and are not used to limit the protection scope of the embodiments of the present application. Any modifications, equivalent substitutions, improvements, etc. made on the basis of the technical solutions of the embodiments of the present application should be included in the protection scope of the embodiments of the present application.

Claims

1. A mobile robot motion planning method for indoor dynamic scenes, characterized in that: include: Generate an indoor environment map and positioning information of the mobile robot based on the environmental perception data acquired in real time, denoise and classify the environmental perception data, and extract static obstacle points and dynamic obstacle points based on recognition features; Construct a comprehensive environmental model of static and dynamic obstacles; Based on the trajectory prediction information of the dynamic obstacle in the comprehensive environment model, a machine learning model is used to identify and mark the current position and predicted position of the dynamic obstacle, thereby identifying the potential key points of the navigation path of the mobile robot; Based on the navigation path key points, a multi-criteria preference function is constructed, and each of the navigation path key points is measured and serialized to obtain a local optimal navigation path key point; Apply the branch and bound method to determine the planning path; Based on the local optimal navigation path key points and travel risk assessment indicators within the planned path range, a guided sampling method is used to generate a planned path to a target location; Converting the planned path into an executable motion trajectory, and finally determining the navigation trajectory of the mobile robot; During the movement of the mobile robot, the planned path and motion trajectory are dynamically adjusted according to the real-time changing environmental perception data.

2. The method for mobile robot motion planning in indoor dynamic scenes according to claim 1, characterized in that: The construction of a comprehensive environment model of static obstacles and dynamic obstacles includes: Building an occupancy grid map of the static environment based on the static obstacle points, marking all inaccessible areas, and forming a static obstacle environment model; Using a motion prediction module to predict the future position and motion trajectory of the dynamic obstacle point to form a dynamic obstacle prediction model; The static obstacle environment model and the dynamic obstacle prediction model are integrated to generate the comprehensive environment model.

3. The method for mobile robot motion planning in indoor dynamic scenes according to claim 1, characterized in that: The construction of a multi-criteria preference function includes: The weighted average method is used to integrate feature importance, movement cost and topological structure to construct a preference function; The preference function is used to calculate the preference value of each navigation path key point, and the navigation path key point with the largest value is configured as the local optimal navigation path key point.

4. The method for mobile robot motion planning in indoor dynamic scenes according to claim 3 is characterized in that: The preference function is: R(b i )=w ig ·IG i -w c ·C i +w topo ·T i (1); Among them, R(b i ) represents the preference function, w ig ,w c ,w topo Respectively represent the weight coefficients of feature importance, movement cost and topological structure; IG i Indicates feature importance; C i represents the moving cost; T i Represents the topological structure.

5. The method for mobile robot motion planning in indoor dynamic scenes according to claim 4, characterized in that: The method for calculating the feature importance includes: Among them, A new (b i ) indicates that the mobile robot moves from the current position to the key point b of the navigation path i After that, the newly explored environment area is A total Indicates that the mobile robot moves to the key point b of the navigation path i The total area that has been explored before, ε is a constant.

6. The method for mobile robot motion planning in indoor dynamic scenes according to claim 4, characterized in that: The calculation method of the movement cost includes: Calculate the mobile robot from the current position p robot Move to candidate boundary point b i The path cost includes: C i =dist(p robot ,b i )(3); Among them, dist(·,·) is the distance function, p robot Indicates the current position, b i It is the key point of the navigation path.

7. The method for mobile robot motion planning in indoor dynamic scenes according to claim 1, characterized in that: The method of generating a planned path to a target location by using a guided sampling method based on the local optimal navigation path key points and the travel risk assessment index within the planned path range includes: The travel risk assessment index is established on the geometric and semantic features of the local point cloud or grid map of the environment to evaluate whether a certain position p is safe for the mobile robot to pass; The travel risk assessment index calculation function is: Among them, G(p) represents the travel risk assessment index, Δz(p) represents the height change within the neighborhood centered on p, to determine whether the ground is flat; h th is the maximum acceptable height difference threshold; α slope (p) is the average slope or maximum slope within the neighborhood, θ th is the passable slope threshold; And / or, a comprehensive determination is made on whether the mobile robot can pass safely by combining normal vector distribution or semantic segmentation results.

8. The method for mobile robot motion planning in indoor dynamic scenes according to claim 7, characterized in that: The method of generating a planned path to a target location by using a guided sampling method based on the local optimal navigation path key points and the travel risk assessment index within the planned path range includes: The key points of the local optimal navigation path and the travel risk assessment index within the planned path range are combined to obtain a guidance set Ψ={(p, G(p)), ...} to provide feasibility guidance for subsequent path planning; When adopting the path planning method based on guided sampling, the points falling in the guided set Ψ are sampled preferentially or forcibly in the sampling space to reduce the exploration of inaccessible or low-access areas, enhance the sampling density of the space near the target key points, and improve the real-time and adaptability of the path.

9. The method for mobile robot motion planning in indoor dynamic scenes according to claim 1, characterized in that: The step of converting the planned path into an executable motion trajectory and finally determining the navigation trajectory of the mobile robot includes: Discretize the generated planned path into several path points, calculate the direction and posture of each path point, and generate the initial executable trajectory by combining the kinematic and dynamic models of the mobile robot; Get the initial executable trace Finally, the objective function J(τ) is constructed by comprehensively considering the safety of environmental obstacles, the smoothness of the trajectory, and the minimum energy consumption of the robot through the optimization method of multi-constraint penalties: Where T is the execution time of the entire trajectory; f obs (x, y) is used to measure the distance cost between the current position (x, y) and the obstacle; κ(t) represents the trajectory curvature, which is represented by Calculate, x′, x″, y′, y″ are the first and second order derivatives of the trajectory with respect to time; a(t) represents the linear acceleration or angular acceleration of the robot, which is expressed in discrete form by |v i+1 -v i | / Δt or |ω i+1 -ω i | / Δt metric; w obs ,w curv ,w acc is the weight coefficient of each penalty; The navigation trajectory is iteratively updated for multiple rounds using gradient descent, heuristic search, or sampling-based optimization methods to gradually reduce the value of J(τ) while satisfying the constraints of velocity, acceleration, and obstacle collision to obtain the final motion planning trajectory.

10. A mobile robot motion planning device for indoor dynamic scenes, characterized in that: A method for mobile robot motion planning in an indoor dynamic scene according to any one of claims 1 to 9 is applied, comprising: Controller (100); A data processing module (200) is electrically connected to the controller (100), and is used to generate an indoor environment map and real-time position information of the mobile robot according to the environmental perception data acquired in real time, to denoise and classify the environmental perception data, and to extract static obstacle points and dynamic obstacle points according to identification features; A model building module (300) is electrically connected to the controller (100), wherein the model building module (300) builds an occupancy grid map of the static environment based on the static obstacle points, marks all inaccessible areas, builds a static obstacle environment model, predicts the future position and motion trajectory of the dynamic obstacle points based on the dynamic obstacle points, and builds a dynamic obstacle prediction model; and finally, the static obstacle environment model and the dynamic obstacle prediction model are integrated to build a comprehensive environment model; A function construction and calculation model (400) is electrically connected to the controller (100), the function construction and calculation model (400) is used to construct a multi-criteria preference function based on the navigation path key points, and measure and serialize each of the navigation path key points to obtain a local optimal navigation path key point; A path planning module (500) is electrically connected to the controller (100), and the path planning module (500) is used to determine the planned path by applying a branch and bound method; and based on the local optimal navigation path key points and the travel risk assessment index within the planned path range, a guided sampling method is used to generate a planned path to a target position; and the path planning module (500) can dynamically adjust the planned path of the mobile robot according to the real-time changing environment and the dynamic obstacle position; A path conversion and optimization module (600) is electrically connected to the controller (100), and is used to convert the planned path into an executable motion trajectory, optimize the navigation trajectory through a multi-constraint optimization algorithm, and finally generate the mobile robot motion planning path by combining environmental safety, trajectory smoothness and energy consumption minimization.

Images