A robot-based path planning method, apparatus, and device

By optimizing state switching rules and improved genetic algorithms, an accurate and efficient robot path plan was generated, solving the problems of model complexity and long time consumption in traditional methods, and realizing flexible path planning for robots in complex environments.

CN117346810BActive Publication Date: 2026-11-10深圳市华赛睿飞智能科技有限公司
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202311354285.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-10-18
Publication Date
2026-11-10
Estimated Expiration
2043-10-18

AI Technical Summary

Technical Problem

Traditional path planning methods are complex in complex environments, with numerous parameters, and tend to converge too early and take too long, resulting in inaccurate path planning.

Method used

The robot state dataset is obtained by adopting state switching rules, a path planning model is constructed, and an improved genetic algorithm is used for optimization, including dynamic mutation rate, global local search and population replacement methods, to generate a robot reference path.

Benefits of technology

It generates accurate and efficient robot reference paths, solving the problems of premature convergence and high time consumption in traditional methods, and enabling the robot to flexibly switch between different motion states.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117346810B_ABST
    Figure CN117346810B_ABST
Patent Text Reader

Abstract

A robot-based path planning method, apparatus, and medium include: acquiring a state dataset of the robot at a preset time step; switching the robot's state according to the state dataset and pre-constructed state switching rules; constructing a robot path planning model corresponding to the robot after the state switch; optimizing a preset genetic algorithm using a preset dynamic mutation rate formula, a global-local search method, and a population replacement method to obtain an improved genetic algorithm; solving the path planning model using the improved genetic algorithm to obtain multiple optimal path points; and performing curve fitting on the multiple optimal path points to obtain a robot reference path. This method addresses the problems of complex models, numerous parameters, premature convergence, and high time consumption in traditional path planning methods, generating an accurate and reliable robot reference path.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and specifically to a robot-based path planning method, apparatus, and device. Background Technology

[0002] With the expansion of the operating environment and application scope of unmanned equipment, single unmanned vehicles or UAVs have limitations in complex indoor application scenarios. This has prompted the development of unmanned aerial vehicles (UAVs) capable of both land and air operations, leveraging the advantages of both. These amphibious robots are generally unmanned, integrated land-air aircraft capable of surviving in both terrestrial and aerial environments. Their ability to move in different media minimizes the complexity of monitoring tasks, enabling data collection in various environments with a single machine. Amphibious robots possess both ground driving and aerial flight modes, and can freely switch between these modes. They can perform aerial tasks by adjusting their flight attitude and also have extended standby time for ground driving tasks. Based on these advantages, amphibious robots can be applied to scenarios such as indoor inspections. During indoor patrols, they can take aerial photos or land on the ground for detailed observation, offering flexibility and efficiency while effectively reducing manpower and financial resources required for patrol operations.

[0003] Robot planning technology is further divided into path planning and motion planning, with path planning being one of the core problems that needs to be solved for amphibious robots to achieve autonomous work. Path planning for amphibious robots refers to generating a smooth, safe path from the starting point to the target point that meets the current work task evaluation criteria, based on certain evaluation standards formed according to the requirements of its work task. The performance of the path planning algorithm directly affects the quality of the amphibious robot's path selection and the smoothness of its movement, and also affects whether the amphibious robot can meet the indicators of minimum energy consumption and maximum speed when performing indoor tasks. Summary of the Invention

[0004] The main technical problem solved by this invention is that path planning modeling is complex with numerous parameters, and traditional path planning methods suffer from premature convergence and high time consumption. This invention overcomes these problems and generates an accurate and reliable robot reference path.

[0005] According to the first aspect, one embodiment provides a robot-based path planning method, comprising:

[0006] Obtain the robot's state dataset at a preset time step, and switch the robot's state according to the state dataset and pre-built state switching rules;

[0007] Construct a robot path planning model corresponding to the robot after the state switch;

[0008] An improved genetic algorithm is obtained by optimizing the preset genetic algorithm using a preset dynamic mutation rate formula, a global-local search method, and a population replacement method.

[0009] The improved genetic algorithm is used to solve the path planning model of the robot to obtain multiple optimal path points. Curve fitting is then performed on the multiple optimal path points to obtain the robot reference path.

[0010] In one embodiment, the construction of the robot path planning model corresponding to the robot after the state switch includes:

[0011] Encoded data is constructed based on the robot's acceleration and angular velocity at different time steps after the state switch;

[0012] A robot path planning model is obtained by using the encoded data and the decoded data constructed based on the encoded data.

[0013] In one embodiment, the step of solving the robot path planning model using the improved genetic algorithm to obtain multiple optimal path points includes:

[0014] Determine whether the reference factor in the dynamic mutation rate formula of the improved genetic algorithm is a preset threshold;

[0015] If the reference factor is a preset threshold, then a global-local search method is used to search in the environment where the robot is located to obtain multiple optimal path points;

[0016] If the reference factor is not equal to the preset threshold, then the dynamic mutation rate corresponding to the dynamic mutation rate formula is calculated;

[0017] Multiple reference path points are obtained using the population replacement method in the improved genetic algorithm, and multiple optimal path points are identified using the dynamic mutation rate and the multiple reference path points.

[0018] In one embodiment, obtaining multiple reference path points according to the population replacement method in the improved genetic algorithm includes:

[0019] Create multiple individuals with varying numbers of individuals, and then aggregate these individuals to obtain an individual population.

[0020] Two individuals are randomly selected from the population, and the parent and parent species of the two individuals are selected.

[0021] Mutation processing is performed on the parent and the parent object using a predefined mutation probability or non-mutation probability, and individuals that meet the preset conditions after mutation processing are selected as multiple reference path points.

[0022] In one embodiment, the step of curve fitting the plurality of optimal path points to obtain a robot reference path includes:

[0023] Several points are selected from the plurality of optimal path points as the endpoints of each piecewise curve, and the path points after removing the endpoints from the plurality of optimal path points are used as the intermediate connection points of each piecewise curve.

[0024] A Bézier curve is fitted based on the endpoints and intermediate connection points to obtain the robot reference path.

[0025] In one embodiment, the step of switching the robot's state according to the state dataset and pre-built state switching rules includes:

[0026] Determine the relationship between the robot's desired point and the preset reference height in the state dataset;

[0027] When the height of the desired point of the robot is greater than the reference height, the state of the robot is switched to flight mode using the state switching rule.

[0028] When the height of the desired point of the robot is less than or equal to the reference height, the state of the robot is switched to ground mode using the state switching rule.

[0029] In one embodiment, before switching the robot's state according to the state dataset and pre-built state switching rules, the method further includes:

[0030] A flight state matrix is ​​constructed based on the robot's flight position and flight attitude angle in flight mode;

[0031] A walking state matrix is ​​constructed based on the robot's walking position and walking posture angle in ground mode;

[0032] Construct a transition matrix between the flight state matrix and the walking state matrix, and use the transition matrix as a pre-constructed state switching rule.

[0033] According to a second aspect, one embodiment provides a robot-based path planning device, comprising:

[0034] The state switching module is used to acquire the state dataset of the robot at a preset time step, and to switch the state of the robot according to the state dataset and the pre-built state switching rules.

[0035] The model building module is used to build the robot path planning model corresponding to the robot after the state switch.

[0036] The algorithm optimization module is used to optimize the preset genetic algorithm using a preset dynamic mutation rate formula, a global-local search method, and a population replacement method to obtain an improved genetic algorithm.

[0037] The path generation module is used to solve the path of the robot path planning model according to the improved genetic algorithm, obtain multiple optimal path points, and perform curve fitting on the multiple optimal path points to obtain the robot reference path.

[0038] The robot-based path planning method, apparatus, and medium according to the above embodiments include switching the robot's state based on a state dataset and pre-built state switching rules. Utilizing these rules enables the robot to flexibly switch between different motion states. A robot path planning model is constructed corresponding to the robot after the state switch, reducing the number of parameters to consider in path planning. An improved genetic algorithm is used to solve the path problem of the robot path planning model, solving the problems of premature convergence and high time consumption inherent in traditional genetic algorithms, and generating an accurate and reliable robot reference path. Attached Figure Description

[0039] Figure 1 This is a flowchart illustrating robot-based path planning in an embodiment of this application.

[0040] Figure 2 This is a schematic diagram of a robot-based path planning process according to one embodiment.

[0041] Figure 3 This is a schematic diagram of a robot-based path planning process according to another embodiment.

[0042] Figure 4 This is a schematic diagram of a robot-based path planning process according to another embodiment.

[0043] Figure 5 This is a schematic diagram of a robot-based path planning process according to another embodiment.

[0044] Figure 6 This is a schematic diagram of a robot-based path planning process according to another embodiment.

[0045] Figure 7 This is a schematic diagram of a robot-based path planning process according to another embodiment.

[0046] Figure 8 This is a structural block diagram of a robot-based path planning device according to an embodiment of this application. Detailed Implementation

[0047] The present invention will now be described in further detail with reference to specific embodiments and accompanying drawings. Similar elements in different embodiments are referred to by associated similar element reference numerals. In the following embodiments, many details are described to facilitate a better understanding of this application. However, those skilled in the art will readily recognize that some features may be omitted in different situations, or may be replaced by other elements, materials, or methods. In some cases, certain operations related to this application are not shown or described in the specification. This is to avoid obscuring the core parts of this application with excessive description. For those skilled in the art, detailed description of these related operations is not necessary; they can fully understand the related operations based on the description in the specification and general technical knowledge in the art.

[0048] Furthermore, the features, operations, or characteristics described in the specification can be combined in any suitable manner to form various embodiments. At the same time, the steps or actions in the method description can be rearranged or adjusted in a manner obvious to those skilled in the art. Therefore, the various orders in the specification and drawings are only for the clear description of a particular embodiment and do not imply a necessary order, unless otherwise stated that a particular order must be followed.

[0049] The serial numbers assigned to components in this document, such as "first" and "second," are used only to distinguish the described objects and have no sequential or technical meaning. The terms "connection" and "linkage" used in this application, unless otherwise specified, include both direct and indirect connections (linkages).

[0050] In some embodiments, the robot's state is switched based on a state dataset and pre-built state switching rules. Utilizing these rules allows the robot to flexibly switch between different motion states. A robot path planning model is constructed corresponding to the robot after the state switch, reducing the number of parameters to consider in path planning. An improved genetic algorithm is then used to solve the path problem of the robot path planning model, resolving the issues of premature convergence and high time consumption inherent in traditional genetic algorithms, and generating an accurate and reliable robot reference path.

[0051] Please refer to Figure 1 Some embodiments of the present invention provide a robot-based path planning method, including steps S10 to S40, which are described in detail below.

[0052] Step S10: Obtain the robot's state dataset at a preset time step, and switch the robot's state according to the state dataset and pre-built state switching rules.

[0053] In some embodiments, the robot's state dataset at a preset time step t is [x t y t , z t v t a t ], where x t and y t The state of the robot at a preset time step t, z t a is the robot's desired point at a preset time step t. t For the robot's acceleration, v t Let t be the robot's angular velocity. t = 1, 2, ..., T, where T is the size of the path planning range.

[0054] See Figure 2 In some embodiments, step S10 includes steps S11 to S13 before switching the robot's state based on the state dataset and pre-built state switching rules, which will be described in detail below.

[0055] Step S11: Construct a flight state matrix based on the robot's flight position and flight attitude angle in flight mode.

[0056] In some embodiments, the robot's flight position in flight mode is (x uav y uav The robot's flight attitude angle in flight mode is θ. uav The constructed flight state matrix is

[0057] Step S12: Construct a walking state matrix based on the robot's walking position and walking posture angle in ground mode.

[0058] In some embodiments, the robot's walking position in ground mode is (x robot y robot The robot's flight attitude angle in ground mode is θ. robot The walking state matrix is

[0059] Step S13: Construct the transition matrix between the flight state matrix and the walking state matrix, and use the transition matrix as a pre-constructed state switching rule.

[0060] In some embodiments, the transition matrix between the flight state matrix and the walking state matrix is ​​constructed as follows: Where RP is the flight state matrix and VP is the walking state matrix. This indicates XOR processing.

[0061] See Figure 3In some embodiments, step S10 includes steps S14 to S16 before switching the robot's state based on the state dataset and pre-built state switching rules, which will be described in detail below.

[0062] Step S14: Determine the relationship between the robot's desired point in the state dataset and the preset reference height.

[0063] In some embodiments, the desired robot point z in the state dataset is determined. t The size of the preset reference height, which can be 1m.

[0064] Step S15: When the height of the desired point of the robot is greater than the reference height, the robot's state is switched to flight mode using the state switching rules.

[0065] In some embodiments, when the robot expects point z t If the height is greater than the reference height by 1m, the robot's state is switched to flight mode using the state switching rules.

[0066] Step S16: When the height of the robot's desired point is less than or equal to the reference height, switch the robot's state to ground mode using the state switching rules.

[0067] In some embodiments, when the robot expects point z t If the height is less than or equal to the reference height of 1m, the robot's state is switched to ground mode using the state switching rules.

[0068] In some embodiments, designing state switching rules and switching states based on the robot's desired points in the state dataset plays an important role in the robot's path planning, enabling the robot to achieve both land operation and air flight modes.

[0069] Step S20: Construct the robot path planning model corresponding to the robot after the state switch.

[0070] See Figure 4 In some embodiments, step S20, which constructs the robot path planning model corresponding to the robot after the state switch, includes steps S21 to S22, which will be described in detail below.

[0071] Step S21: Construct encoded data based on the robot's acceleration and angular velocity at different time steps after the state switch.

[0072] Step S22: Use the encoded data and the decoded data constructed based on the encoded data to build a model and obtain the robot path planning model.

[0073] In some embodiments, the encoding DNA is represented as DNA = [gene1, gene2, ..., gene] t ,...,gene T ], encoded data gene t =(a t ,e t ), a t For the robot's acceleration, v t Let T be the robot's angular velocity. The encoded DNA is defined as a set of drone planners, where T is the size of the path planning range.

[0074] In some embodiments, the decoded data is constructed based on the encoded data, therefore the decoded DNA is γDNA = [γgene1, γgene2, ..., γgene]. t ,...,γgene T ], γgene t =[x t y t , z t v t a t ], where γ is the preset decoding parameter.

[0075] In some embodiments, using encoded DNA and decoded DNA to establish a robot path planning model can simplify the robot path planning model, reduce the number of parameters that need to be considered, and greatly improve the efficiency of the algorithm.

[0076] Step S30: Optimize the preset genetic algorithm using the preset dynamic mutation rate formula, global-local search method, and population replacement method to obtain an improved genetic algorithm.

[0077] In some embodiments, solving the robot path planning model using a traditional genetic algorithm may result in premature convergence and high fitness time. To address these issues, the preset genetic algorithm can be optimized using a preset dynamic mutation rate formula, a global-local search method, and a population replacement method to obtain an improved genetic algorithm.

[0078] Step S40: Solve the path planning model of the robot using the improved genetic algorithm to obtain multiple optimal path points. Perform curve fitting on the multiple optimal path points to obtain the robot reference path.

[0079] See Figure 5 In some embodiments, step S40 solves the path planning model of the robot according to the improved genetic algorithm to obtain multiple optimal path points, including steps S41 to S44, which are described in detail below.

[0080] Step S41: Determine whether the reference factor in the dynamic mutation rate formula of the improved genetic algorithm is a preset threshold.

[0081] In some embodiments, the reference factor in the dynamic mutation rate formula of the improved genetic algorithm is the f-factor, where the f-factor includes the distance f. dist Obstacle position f obs Segmented angles and f curv and battery power generation function f cons . f dist The distance from the current point to the target point. This represents the coordinates of the current point. The coordinates of the target point are represented by T, the size of the path planning range is T, γ is the preset decoding parameter, d is the preset coordinate parameter, and (x, y) is the coordinate of the robot.

[0082] In some embodiments, the distance f dist If it is less than the threshold ε, then set it to 0, and the obstacle position f obs If there are no obstacles in the middle, set it to 0, segment angle and f curv It is the sum of angles between each segment of the route, and the battery charge generation function f. cons Battery power generation has been taken into account.

[0083] Step S42: If the reference factor is a preset threshold, then the global-local search method is used to search in the robot's environment to obtain multiple optimal path points.

[0084] Step S43: If the reference factor is not equal to the preset threshold, calculate the dynamic variation rate corresponding to the dynamic variation rate formula.

[0085] Step S44: Obtain multiple reference path points according to the population replacement method in the improved genetic algorithm, and identify multiple optimal path points using the dynamic mutation rate and multiple reference path points.

[0086] See Figure 6 In some embodiments, step S44 obtains multiple reference path points according to the population replacement method in the improved genetic algorithm, including steps S441 to S443, which are described in detail below.

[0087] Step S441: Create multiple individuals with different numbers of individuals, and obtain an individual population by summing up the multiple individuals.

[0088] Step S442: Randomly select two individuals from the individual population and filter out the parent and parent object from the two individuals.

[0089] In some embodiments, a tournament-style selection and screening method is used.

[0090] Step S443: Perform mutation processing on the parent and parent objects using predefined mutation probabilities or non-mutation probabilities, and select individuals that meet the preset conditions after mutation processing as multiple reference path points.

[0091] In some embodiments, a mutation operator is performed on each new gene with a probability of mutation or non-mutation. Four operators for performing mutation are defined: creep, change, insertion, and removal. Creep adds or subtracts a small random value from the elements of each gene; change has a 50% probability that each gene will be replaced by a randomly generated new gene; insertion refers to... At time, a single randomly generated gene is inserted into any location on the DNA; removal refers to... At that time, random genes are removed from the DNA. Among them, For time step, To preset the maximum time step, This is the preset minimum time step.

[0092] See Figure 7 In some embodiments, step S40 performs curve fitting on multiple optimal path points to obtain the robot reference path, including steps S444 to S445, which are described in detail below.

[0093] Step S444: Select several points from multiple optimal path points as the endpoints of each piecewise curve, and use the path points after removing the endpoints from the multiple optimal path points as the intermediate connection points of each piecewise curve.

[0094] Step S445: Fit a Bézier curve based on the endpoints and intermediate connection points to obtain the robot reference path.

[0095] In some embodiments, a robot reference path is obtained by performing piecewise cubic Bézier curve fitting on a continuous curve based on multiple optimal path points.

[0096] In some embodiments, traditional genetic algorithms suffer from premature convergence and high fitness time. Therefore, an improved genetic algorithm is used to solve the problem, providing a dynamic mutation rate and a switchable global-local search method for the mutation operator of the traditional genetic algorithm.

[0097] The robot-based path planning method, apparatus, and medium according to the above embodiments include switching the robot's state based on a state dataset and pre-built state switching rules. Utilizing these rules enables the robot to flexibly switch between different motion states. A robot path planning model corresponding to the robot after the state switch is constructed, reducing the number of parameters to consider in path planning. An improved genetic algorithm is used to solve the path problem of the robot path planning model, solving the problems of premature convergence and high time consumption inherent in traditional genetic algorithms, and generating an accurate and reliable robot reference path.

[0098] Please refer to Figure 8 Some embodiments provide a robot-based path planning device, including a state switching module 10, a model building module 20, an algorithm optimization module 30, and a path generation module 40, which are described in detail below.

[0099] The state switching module 10 is used to acquire the robot's state dataset at a preset time step and switch the robot's state according to the state dataset and pre-built state switching rules.

[0100] In some embodiments, the robot's state dataset at a preset time step t is [x t y t , z t v t a t ], where x t and y t The state of the robot at a preset time step t, z t a is the robot's desired point at a preset time step t. t For the robot's acceleration, v t Let t be the robot's angular velocity. t = 1, 2, ..., T, where T is the size of the path planning range.

[0101] Please return to the previous page. Figure 2 In some embodiments, the state switching module 10 may also perform the following actions before switching the robot's state according to the state dataset and pre-built state switching rules, which are described in detail below.

[0102] The state switching module 10 constructs a flight state matrix based on the robot's flight position and flight attitude angle in flight mode.

[0103] In some embodiments, the robot's flight position in flight mode is (x uav y uav The robot's flight attitude angle in flight mode is θ. uav The constructed flight state matrix is

[0104] The state switching module 10 constructs a walking state matrix based on the robot's walking position and walking posture angle in ground mode.

[0105] In some embodiments, the robot's walking position in ground mode is (x robot y robot The robot's flight attitude angle in ground mode is θ. robot The walking state matrix is

[0106] The state transition module 10 constructs the transition matrix between the flight state matrix and the walking state matrix, and uses the transition matrix as a pre-constructed state transition rule.

[0107] In some embodiments, the transition matrix between the flight state matrix and the walking state matrix is ​​constructed as follows: VP. Where RP is the flight state matrix and VP is the walking state matrix. This indicates XOR processing.

[0108] Please return to the previous page. Figure 3 In some embodiments, the state switching module 10 may also perform the following actions before switching the robot's state according to the state dataset and pre-built state switching rules, which are described in detail below.

[0109] The state switching module 10 determines the relationship between the robot's desired point in the state dataset and the preset reference height.

[0110] In some embodiments, the desired robot point z in the state dataset is determined. t The size of the preset reference height, which can be 1m.

[0111] When the desired height of the robot is greater than the reference height, the state switching module 10 switches the robot's state to flight mode using the state switching rules.

[0112] In some embodiments, when the robot expects point z t If the height is greater than the reference height by 1m, the robot's state is switched to flight mode using the state switching rules.

[0113] When the height of the desired point of the robot is less than or equal to the reference height, the state switching module 10 switches the robot's state to ground mode using the state switching rules.

[0114] In some embodiments, when the robot expects point z t If the height is less than or equal to the reference height of 1m, the robot's state is switched to ground mode using the state switching rules.

[0115] In some embodiments, designing state switching rules and switching states based on the robot's desired points in the state dataset plays an important role in the robot's path planning, enabling the robot to achieve both land operation and air flight modes.

[0116] The model building module 20 is used to build the robot path planning model corresponding to the robot after the state switch.

[0117] Please return to the previous page. Figure 4 In some embodiments, the model building module 20 is used to build a robot path planning model corresponding to the robot after the state switch, and can perform the following actions, which are described in detail below.

[0118] The model building module 20 constructs encoded data based on the robot's acceleration and angular velocity at different time steps after the state switch. It then uses the encoded data and the decoded data constructed based on the encoded data to build the model and obtain the robot path planning model.

[0119] In some embodiments, the encoding DNA is represented as DNA = [gene1, gene2, ..., gene] t ,...,gene T ], encoded data gene t =(a t ,e t ), a t For the robot's acceleration, v t Let T be the robot's angular velocity. The encoded DNA is defined as a set of drone planners, where T is the size of the path planning range.

[0120] In some embodiments, the decoded data is constructed based on the encoded data, therefore the decoded DNA is γDNA = [γgene1, γgene2, ..., γgene]. t ,...,γgene T ], γgene t =[x t y t , z t v t a t ], where γ is the preset decoding parameter.

[0121] In some embodiments, using encoded DNA and decoded DNA to establish a robot path planning model can simplify the robot path planning model, reduce the number of parameters that need to be considered, and greatly improve the efficiency of the algorithm.

[0122] The algorithm optimization module 30 is used to optimize the preset genetic algorithm using a preset dynamic mutation rate formula, a global local search method, and a population replacement method to obtain an improved genetic algorithm.

[0123] In some embodiments, solving the robot path planning model using a traditional genetic algorithm may result in premature convergence and high fitness time. To address these issues, the preset genetic algorithm can be optimized using a preset dynamic mutation rate formula, a global-local search method, and a population replacement method to obtain an improved genetic algorithm.

[0124] The path generation module 40 is used to solve the path planning model of the robot according to the improved genetic algorithm, obtain multiple optimal path points, and perform curve fitting on the multiple optimal path points to obtain the robot reference path.

[0125] Please return to the previous page. Figure 5 In some embodiments, the path generation module 40 is used to solve the path of the robot path planning model according to the improved genetic algorithm and obtain multiple optimal path points. The following actions can be performed, which are described in detail below.

[0126] The path generation module 40 determines whether the reference factor in the dynamic mutation rate formula of the improved genetic algorithm is a preset threshold. If the reference factor is a preset threshold, a global-local search method is used to search in the robot's environment to obtain multiple optimal path points. If the reference factor is not equal to the preset threshold, the dynamic mutation rate corresponding to the dynamic mutation rate formula is calculated, and multiple reference path points are obtained according to the population replacement method in the improved genetic algorithm. Multiple optimal path points are then identified using the dynamic mutation rate and the multiple reference path points.

[0127] In some embodiments, the reference factor in the dynamic mutation rate formula of the improved genetic algorithm is the f-factor, where the f-factor includes the distance f. dist Obstacle position f obs Segmented angles and f curv and battery power generation function f cons . f dist The distance from the current point to the target point. This represents the coordinates of the current point. The coordinates of the target point are represented by T, the size of the path planning range is T, γ is the preset decoding parameter, d is the preset coordinate parameter, and (x, y) is the coordinate of the robot.

[0128] In some embodiments, the distance f dist If it is less than the threshold ε, then set it to 0, and the obstacle position f obs If there are no obstacles in the middle, set it to 0, segment angle and f curv It is the sum of angles between each segment of the route, and the battery charge generation function f. cons Battery power generation has been taken into account.

[0129] Please return to the previous page. Figure 6 In some embodiments, the path generation module 40 can perform the following actions to obtain multiple reference path points according to the population replacement method in the improved genetic algorithm, which are described in detail below.

[0130] The path generation module 40 creates multiple individuals with varying numbers of individuals, summarizes these individuals to obtain an individual population, randomly selects two individuals from the individual population, and filters out the parent and parent objects of these two individuals.

[0131] In some embodiments, a tournament-style selection and screening method is used.

[0132] The path generation module 40 performs mutation processing on the parent and parent objects using predefined mutation probabilities or non-mutation probabilities, and selects individuals that meet preset conditions after mutation processing as multiple reference path points.

[0133] In some embodiments, a mutation operator is performed on each new gene with a probability of mutation or non-mutation. Four operators for performing mutation are defined: creep, change, insertion, and removal. Creep adds or subtracts a small random value from the elements of each gene; change has a 50% probability that each gene will be replaced by a randomly generated new gene; insertion refers to... At time, a single randomly generated gene is inserted into any location on the DNA; removal refers to... At that time, random genes are removed from the DNA. Among them, For time step, To preset the maximum time step, This is the preset minimum time step.

[0134] Please return to the previous page. Figure 7 In some embodiments, the path generation module 40 is used to perform curve fitting on multiple optimal path points to obtain a robot reference path, which can perform the following actions, as described in detail below.

[0135] The path generation module 40 selects several points from multiple optimal path points as the endpoints of each segmented curve, and the path points after removing the endpoints from the multiple optimal path points are used as the intermediate connection points of each segmented curve. Based on the endpoints and intermediate connection points, a Bézier curve is fitted to obtain the robot reference path.

[0136] In some embodiments, a robot reference path is obtained by performing piecewise cubic Bézier curve fitting on a continuous curve based on multiple optimal path points.

[0137] In some embodiments, traditional genetic algorithms suffer from premature convergence and high fitness time. Therefore, an improved genetic algorithm is used to solve the problem, providing a dynamic mutation rate and a switchable global-local search method for the mutation operator of the traditional genetic algorithm.

[0138] The robot-based path planning method, apparatus, and medium according to the above embodiments include switching the robot's state based on a state dataset and pre-built state switching rules. Utilizing these rules enables the robot to flexibly switch between different motion states. A robot path planning model corresponding to the robot after the state switch is constructed, reducing the number of parameters to consider in path planning. An improved genetic algorithm is used to solve the path problem of the robot path planning model, solving the problems of premature convergence and high time consumption inherent in traditional genetic algorithms, and generating an accurate and reliable robot reference path.

[0139] Those skilled in the art will understand that all or part of the functions of the various methods in the above embodiments can be implemented by hardware or by computer programs. When all or part of the functions in the above embodiments are implemented by computer programs, the program can be stored in a computer-readable storage medium, which may include: read-only memory, random access memory, disk, optical disk, hard disk, etc., and the program is executed by a computer to achieve the above functions. For example, the program can be stored in the memory of a device, and when the program in the memory is executed by the processor, all or part of the above functions can be achieved. In addition, when all or part of the functions in the above embodiments are implemented by computer programs, the program can also be stored in a server, another computer, disk, optical disk, flash drive, or external hard drive, etc., and can be downloaded or copied to the memory of a local device, or the system of the local device can be updated. When the program in the memory is executed by the processor, all or part of the functions in the above embodiments can be achieved.

[0140] The above examples illustrate the present invention only to aid in understanding it and are not intended to limit the scope of the invention. Those skilled in the art can make various simple deductions, modifications, or substitutions based on the principles of this invention.

Claims

1. A robot-based path planning method, characterized in that, include: Obtain the robot's state dataset at a preset time step, and switch the robot's state according to the state dataset and pre-built state switching rules; Construct a robot path planning model corresponding to the robot after the state switch; An improved genetic algorithm is obtained by optimizing the preset genetic algorithm using a preset dynamic mutation rate formula, a global-local search method, and a population replacement method. Determine whether the reference factor in the dynamic mutation rate formula of the improved genetic algorithm is a preset threshold; if the reference factor is a preset threshold, then use a global-local search method to search in the robot's environment to obtain multiple optimal path points; if the reference factor is not equal to the preset threshold, then calculate the dynamic mutation rate corresponding to the dynamic mutation rate formula; create multiple individuals with different numbers, and summarize the multiple individuals to obtain an individual population; randomly select two individuals from the individual population, and filter out the parent and parent objects from the two individuals; perform mutation processing on the parent and parent objects using a predefined mutation probability or non-mutation probability, and select individuals that meet the preset conditions after mutation processing as multiple reference path points, and use the dynamic mutation rate and the multiple reference path points to identify multiple optimal path points; perform curve fitting on the multiple optimal path points to obtain the robot reference path.

2. The method as described in claim 1, characterized in that, The robot path planning model corresponding to the robot after the state switch is constructed includes: Encoded data is constructed based on the robot's acceleration and angular velocity at different time steps after the state switch; A robot path planning model is obtained by using the encoded data and the decoded data constructed based on the encoded data.

3. The method as described in claim 1, characterized in that, The step of performing curve fitting on the multiple optimal path points to obtain the robot reference path includes: Several points are selected from the plurality of optimal path points as the endpoints of each piecewise curve, and the path points after removing the endpoints from the plurality of optimal path points are used as the intermediate connection points of each piecewise curve. A Bézier curve is fitted based on the endpoints and intermediate connection points to obtain the robot reference path.

4. The method as described in claim 1, characterized in that, The step of switching the robot's state according to the state dataset and pre-built state switching rules includes: Determine the relationship between the robot's desired point and the preset reference height in the state dataset; When the height of the desired point of the robot is greater than the reference height, the state of the robot is switched to flight mode using the state switching rule. When the height of the desired point of the robot is less than or equal to the reference height, the state of the robot is switched to ground mode using the state switching rule.

5. The method as described in claim 1, characterized in that, Before switching the robot's state based on the state dataset and pre-built state switching rules, the process also includes: A flight state matrix is ​​constructed based on the robot's flight position and flight attitude angle in flight mode; A walking state matrix is ​​constructed based on the robot's walking position and walking posture angle in ground mode; Construct a transition matrix between the flight state matrix and the walking state matrix, and use the transition matrix as a pre-constructed state switching rule.

6. A robot-based path planning device, characterized in that, include: The state switching module is used to acquire the state dataset of the robot at a preset time step, and to switch the state of the robot according to the state dataset and the pre-built state switching rules. The model building module is used to build the robot path planning model corresponding to the robot after the state switch. The algorithm optimization module is used to optimize the preset genetic algorithm using a preset dynamic mutation rate formula, a global-local search method, and a population replacement method to obtain an improved genetic algorithm. The path generation module is used to determine whether the reference factor in the dynamic mutation rate formula of the improved genetic algorithm is a preset threshold. If the reference factor is a preset threshold, a global-local search method is used to search the environment where the robot is located to obtain multiple optimal path points. If the reference factor is not equal to the preset threshold, the dynamic mutation rate corresponding to the dynamic mutation rate formula is calculated. Multiple individuals with different numbers are created, and an individual population is obtained by summing the multiple individuals. Two individuals are randomly selected from the individual population, and the parent and parent objects of the two individuals are screened out. Mutation processing is performed on the parent and parent objects using a predefined mutation probability or non-mutation probability, and individuals that meet the preset conditions after mutation processing are selected as multiple reference path points. Multiple optimal path points are identified using the dynamic mutation rate and the multiple reference path points. Curve fitting is performed on the multiple optimal path points to obtain the robot reference path.

7. A robot-based path planning device, characterized in that, include: Memory, used to store programs; A processor for implementing the method as described in any one of claims 1-5 by executing a program stored in the memory.

8. A computer-readable storage medium, characterized in that, The medium stores a program that can be executed by a processor to implement the method as described in any one of claims 1-5.

Citation Information

Patent Citations

  • Mobile robot path planning method and system based on hybrid improved ant colony algorithm

    CN115562273A

  • Motion path planning method of unmanned aerial vehicle, unmanned aerial vehicle and storage medium

    CN116859979A