CN114518758B - Indoor measurement robot multi-target point moving path planning method based on Q learning - Google Patents
Indoor measurement robot multi-target point moving path planning method based on Q learning Download PDFInfo
- Publication number
- CN114518758B CN114518758B CN202210118037.9A CN202210118037A CN114518758B CN 114518758 B CN114518758 B CN 114518758B CN 202210118037 A CN202210118037 A CN 202210118037A CN 114518758 B CN114518758 B CN 114518758B
- Authority
- CN
- China
- Prior art keywords
- measuring
- target point
- robot
- paths
- return
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 20
- 238000005259 measurement Methods 0.000 title claims description 49
- 230000006870 function Effects 0.000 claims abstract description 21
- 238000010586 diagram Methods 0.000 claims description 6
- 239000011159 matrix material Substances 0.000 claims description 6
- 238000004891 communication Methods 0.000 claims description 4
- 230000003111 delayed effect Effects 0.000 claims description 4
- 230000007613 environmental effect Effects 0.000 claims description 3
- 230000002787 reinforcement Effects 0.000 abstract description 2
- 238000010276 construction Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 238000004904 shortening Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
- G05D1/02—Control of position or course in two dimensions
- G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
- G05D1/0212—Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory
- G05D1/0217—Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory in accordance with energy consumption, time reduction or distance reduction criteria
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
- G05D1/02—Control of position or course in two dimensions
- G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
- G05D1/0212—Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory
- G05D1/0221—Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory involving a learning process
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Landscapes
- Engineering & Computer Science (AREA)
- Aviation & Aerospace Engineering (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Manipulator (AREA)
- Feedback Control In General (AREA)
Abstract
The invention provides a method for planning a multi-target point moving path of an indoor measuring robot based on Q learning, which is based on a model-independent Q learning algorithm in reinforcement learning, wherein a moving strategy conforming to the measuring robot is set in a return function for robot practice learning, three elements of states, actions and rewards are utilized, actions are taken according to the current state of the measuring robot, the rewards which are fed back are recorded, and better actions which can be taken when the measuring robot is in the next state are decided, so that the measuring robot automatically acquires a moving path with reasonable sequence under the condition that the number and the position of measuring points are known. The moving path planning method provided by the invention can realize global moving planning of multiple measuring points of the measuring robot in an indoor environment, help the measuring robot to complete work at the fastest speed without no need in the process of carrying out a measuring task, and greatly improve the measuring efficiency.
Description
Technical Field
The invention belongs to the technical field of mobile robots, and particularly relates to a multi-target point moving path planning method of an indoor measurement robot based on Q learning.
Background
In traditional house measurement work, need a large amount of manpower to intervene, work load is heavy, in addition, can also lead to producing more measurement errors because of the human factor in the measurement process, is difficult to guarantee the accuracy of measurement. The robot is used for automatic measurement, so that the measurement precision and the working efficiency can be effectively improved, the workload of measurement staff and excessive labor expenditure are reduced, and the aims of reducing the cost and shortening the construction period can be achieved.
In an indoor environment, the measuring robots are required to enter each room one by one to finish the measuring task, the moving paths of the existing measuring robots are selected by personnel manually in sequence, are ordered according to the selecting sequence and move according to the sequence, so that the moving paths cannot be guaranteed to be most efficient, the construction efficiency is not improved, and the method is not applicable after the robots automatically identify multiple measuring points at one time according to a map. Therefore, in order to realize that the robot passes through all measuring points in the process of going in and going out, and the moving process does not make an unnecessary way, a most efficient path needs to be reasonably planned.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a multi-target point moving path planning method for an indoor measuring robot based on Q learning, which effectively solves the problems of moving target point sequencing and moving planning of the measuring robot in the indoor measuring process and greatly improves the measuring efficiency.
The present invention achieves the above technical object by the following means.
The method for planning the movement path of the multiple target points of the indoor measurement robot based on Q learning comprises the following steps:
step 1: the measuring robot identifies all measuring target points according to the house type graph, wherein each room is a measuring point and marked by a digital serial number, and a path schematic diagram is obtained according to the positions of the measuring target points and the communication mode among the measuring target points;
step 2: setting a return function, taking the direction facing the entrance of the robot as a reference direction, and defining the direction of the starting point facing the measurement target point as a positive direction if the included angle between the direction of the starting point facing the measurement target point and the reference direction is smaller than or equal to 90 degrees, and defining the direction as a negative direction if the included angle is smaller than or equal to 90 degrees; all the measurement target points are connected according to actual paths, wherein the paths with the same route and opposite directions are two positive and negative paths, and the distance between the paths corresponding to the positive direction is a positive distance, and the distance between the paths is a negative distance; setting a set of measurement target points as a state space S, setting a set of all paths as an action space A, setting a path return value rewards of each measurement target point, and acquiring a return matrix matched with the house type graph according to a return function;
step 3: the measuring robot explores in a way of taking an action from one state to another until reaching a target state; taking each exploration as one experience, wherein each experience comprises a process of measuring the robot to reach a target state from an initial state, and turning to the next exploration after the measuring robot reaches the target state each time;
constructing a two-dimensional robot memory table Q_table according to all states and actions experienced by the measuring robot, wherein the initial state of the Q_table is 0, under the condition that the environment is unknown, the Q_table is initially set to be only one element, and if a new state is found, the Q_table is added with a new row and a new column;
step 4: defining a Q function shown in the following formula (1) as a utility value function of the robot action for updating the Q_table:
Q(s,a)=R(s,a)+γV * ;
wherein S is one of the states of S; a is one action in A; r is a return function, and is used for calculating an immediate return value obtained by the measuring robot after selecting actions from the current state and transferring to the next state, and the immediate return value is determined by the current state and the selected actions; gamma is a discount factor for determining the relative proportion of delayed return to immediate return; v is the maximum return in all actions in the current state, V * =max[R(s,A)];
Step 5: updating the Q_table according to the following formula to obtain an optimal moving path:
Q(s t ,a t )←R(s t ,a t )+γ*max[Q(s t+1 ,A t+1 )]
where t represents the current time.
Further, the return function satisfies a multi-target point efficient movement strategy of the measuring robot, namely, the already-moved measuring target point is not used as the measuring target point any more, and the path originally divided by the measuring target point becomes a continuous path.
Further, the path reward value rewards of each measurement target point are set as follows:
the rewarding value satisfying the shortest positive distance between the paths of each measuring target point is 70;
the prize value satisfying the positive direction between the paths of each measurement target point is 90;
the rewarding value which satisfies the positive direction and has the shortest positive distance between the paths of each measuring target point is 110;
the shortest reward value satisfying the inverse distance between the paths of each measurement target point is 10;
the reward value satisfying the opposite direction between the paths of each measurement target point is 30;
the reward value which satisfies the opposite direction and has the shortest opposite distance between the paths of each measuring target point is 50;
the prize value for which all conditions are not met is 0.
Further, the step of updating the q_table is as follows:
s1: setting a parameter gamma and an environmental reward value in a return matrix;
s2: initializing Q_table to 0;
s3: determining an initial state of the measuring robot;
s4: selecting one possible action from all possible actions of the current state;
s5: using this possible action, the measuring robot reaches the next state;
s6: for the next state, obtaining the maximum Q value based on all possible actions thereof;
s7: the Q value before the replacement is replaced according to the Q_table updating formula;
s8: repeating the steps S4 to S7 until the Q_table converges.
The invention has the following beneficial effects:
the invention utilizes a model-independent Q learning algorithm in reinforcement learning, sets a movement strategy conforming to the measuring robot in a return function for robot practice learning based on the learning capability of Q learning, takes action according to the current state of the measuring robot by using three elements of state (state), action and reward (reward), records the rewards fed back, and decides a better action which can be taken when the measuring robot is in the next state, thereby realizing that the measuring robot automatically acquires a movement path with reasonable sequence, namely a path with shortest movement time and least repeated route under the condition of known number and position of measuring points. The moving path planning method provided by the invention can realize global moving planning of multiple measuring points of the measuring robot in an indoor environment, help the measuring robot to complete work at the fastest speed without no need in the process of carrying out a measuring task, and greatly improve the measuring efficiency.
Drawings
FIG. 1 is a schematic diagram of measurement target points identified from a house pattern;
FIG. 2 is a schematic diagram showing the communication between measurement target points;
FIG. 3 is a flow chart of robot memory table training.
Detailed Description
The invention will be further described with reference to the drawings and the specific embodiments, but the scope of the invention is not limited thereto.
The invention relates to a method for planning a multi-target point moving path of an indoor measurement robot based on Q learning, which comprises the following steps:
step 1: the measuring robot identifies all measuring target points shown in figure 1 according to the house type diagram, wherein each room is a measuring point and is marked by a digital serial number; according to the positions of the measurement target points and the communication mode between the measurement target points, simplifying the process of fig. 1 into a path schematic diagram shown in fig. 2 (the numerical marks in fig. 1 and 2 represent room marks);
step 2: setting a return function, wherein the return function meets a multi-target point efficient movement strategy of the measuring robot, namely the measurement target point which is already moved is not used as the measurement target point any more, so that a path originally divided by the measurement target point becomes a continuous path; the setting strategy of the reporting function is as follows: taking the direction facing the robot entrance as a reference direction, if the included angle between the direction facing the measurement target point of the starting point and the reference direction is smaller than or equal to 90 degrees, defining the direction as a positive direction, and otherwise defining the direction as a negative direction; all the measurement target points are connected according to the actual paths (the paths with the same route and opposite directions are two positive and negative paths), and the distance between the paths corresponding to the positive direction is a positive distance, and the distance between the paths is a negative distance;
setting a set of measurement target points as a state space S, a set of all paths as an action space a, and setting a path return value reward for each measurement target point as follows:
the rewarding value satisfying the shortest positive distance between the paths of each measuring target point is 70;
the prize value satisfying the positive direction between the paths of each measurement target point is 90;
the rewarding value which satisfies the positive direction and has the shortest positive distance between the paths of each measuring target point is 110;
the shortest reward value satisfying the inverse distance between the paths of each measurement target point is 10;
the reward value satisfying the opposite direction between the paths of each measurement target point is 30;
the reward value which satisfies the opposite direction and has the shortest opposite distance between the paths of each measuring target point is 50;
a prize value of 0 that is not satisfied by all conditions;
and obtaining a return matrix matched with the house type graph according to the return function.
Step 3: the measuring robot searches from one state (state) to another state (state) until reaching the target state; taking each exploration as one experience, wherein each experience comprises the process that the measuring robot reaches a target state (namely the last measuring target point position) from an initial state (namely the starting point position), and turning to the next exploration after each measuring robot reaches the target state;
a two-dimensional robot memory table Q_table is constructed according to all states and actions experienced by the measuring robot, the initial state of the Q_table is 0, under the condition that the environment is unknown, the Q_table is initially set to be only one element, and if a new state is found, the Q_table is added with a new row and a new column.
Step 4: defining a Q function shown in the following formula (1) as a utility value function of the robot action for updating the Q_table:
Q(s,a)=R(s,a)+γV * (1)
wherein S is a discrete bounded state space, S is a state in S; a is a discrete action space, and a is one action in A; r is a return function, and is used for calculating an immediate return value obtained by the measuring robot after selecting actions from the current state and transferring to the next state, and the immediate return value is determined by the current state and the selected actions; gamma is a discount factor for determining the relative proportion of delayed return to immediate return, with a larger gamma value indicating a greater importance of delayed return; v is the maximum return in all actions in the current state, V * =max[R(s,A)]。
Step 5: the Q_table is updated according to the following formula (2):
Q(s t ,a t )←R(s t ,a t )+γ*max[Q(s t+1 ,A t+1 )] (2)
wherein t represents the current time;
updating the Q_table can enhance the judging capability of the measuring robot when moving, and more updated results lead to better Q_table; in this case, if the q_table has been enhanced, the measurement robot will find the most appropriate target point movement sequence, i.e., the optimal movement path;
the update flow is shown in fig. 3, and the specific update steps are as follows:
s1: setting a parameter gamma and an environmental reward value in a return matrix;
s2: initializing Q_table to 0;
s3: determining an initial state of the measuring robot;
s4: selecting one possible action from all possible actions of the current state;
s5: using this possible action, the measuring robot reaches the next state;
s6: for the next state, obtaining the maximum Q value based on all possible actions thereof;
s7: replacing the previous Q value according to the Q_table updating mode;
s8: repeating the steps S4 to S7 until the Q_table converges.
The examples are preferred embodiments of the present invention, but the present invention is not limited to the above-described embodiments, and any obvious modifications, substitutions or variations that can be made by one skilled in the art without departing from the spirit of the present invention are within the scope of the present invention.
Claims (1)
1. The method for planning the multi-target point moving path of the indoor measurement robot based on Q learning is characterized by comprising the following steps:
step 1: the measuring robot identifies all measuring target points according to the house type graph, and obtains a path schematic diagram according to the positions of the measuring target points and the communication mode among the measuring target points;
step 2: setting a return function, taking the direction facing the entrance of the robot as a reference direction, and defining the direction of the starting point facing the measurement target point as a positive direction if the included angle between the direction of the starting point facing the measurement target point and the reference direction is smaller than or equal to 90 degrees, and defining the direction as a negative direction if the included angle is smaller than or equal to 90 degrees; all the measurement target points are connected according to actual paths, wherein the paths with the same route and opposite directions are two positive and negative paths, and the distance between the paths corresponding to the positive direction is a positive distance, and the distance between the paths is a negative distance; setting a set of measurement target points as a state space S, setting a set of all paths as an action space A, setting a path return value rewards of each measurement target point, and acquiring a return matrix matched with the house type graph according to a return function;
step 3: the measuring robot searches in a mode of taking one action from one state to another state until reaching a target state, and then switches to search for the next time; constructing a two-dimensional robot memory table Q_table according to all states and actions experienced by the measuring robot, wherein the initial state of the Q_table is 0, the Q_table initially has only one element, and after a new state is found, the Q_table is added with a new row and a new column;
step 4: defining a Q function as a utility function of the robot action for updating the Q_table:
Q(s,a)=R(s,a)+γV * ;
wherein S is one of the states of S; a is one action in A; r is a return function, and is used for calculating an immediate return value obtained by the measuring robot after selecting actions from the current state and transferring to the next state, and the immediate return value is determined by the current state and the selected actions; gamma is a discount factor for determining the relative proportion of delayed return to immediate return; v is the maximum return in all actions in the current state, V * =max[R(s,A)];
Step 5: updating the Q_table according to the following formula to obtain an optimal moving path:
Q(s t ,a t )←R(s t ,a t )+γ*max[Q(s t+1 ,A t+1 )]
wherein t represents the current time;
the return function meets the efficient movement strategy of multiple target points of the measuring robot, namely the measurement target point which is already moved is not used as the measurement target point any more, and the path originally divided by the measurement target point becomes a continuous path;
the path return value rewards of each measurement target point are set as follows:
the rewarding value satisfying the shortest positive distance between the paths of each measuring target point is 70;
the prize value satisfying the positive direction between the paths of each measurement target point is 90;
the rewarding value which satisfies the positive direction and has the shortest positive distance between the paths of each measuring target point is 110; the shortest reward value satisfying the inverse distance between the paths of each measurement target point is 10;
the reward value satisfying the opposite direction between the paths of each measurement target point is 30;
the reward value which satisfies the opposite direction and has the shortest opposite distance between the paths of each measuring target point is 50; a prize value of 0 that is not satisfied by all conditions;
the Q_table updating step is as follows:
s1: setting a parameter gamma and an environmental reward value in a return matrix;
s2: initializing Q_table to 0;
s3: determining an initial state of the measuring robot;
s4: selecting one possible action from all possible actions of the current state;
s5: the measuring robot reaches the next state using the possible actions determined in S4;
s6: for the next state, obtaining the maximum Q value based on all possible actions thereof;
s7: the Q value before the replacement is replaced according to the Q_table updating formula;
s8: repeating the steps S4 to S7 until the Q_table converges.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210118037.9A CN114518758B (en) | 2022-02-08 | 2022-02-08 | Indoor measurement robot multi-target point moving path planning method based on Q learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210118037.9A CN114518758B (en) | 2022-02-08 | 2022-02-08 | Indoor measurement robot multi-target point moving path planning method based on Q learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114518758A CN114518758A (en) | 2022-05-20 |
CN114518758B true CN114518758B (en) | 2023-12-12 |
Family
ID=81596759
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210118037.9A Active CN114518758B (en) | 2022-02-08 | 2022-02-08 | Indoor measurement robot multi-target point moving path planning method based on Q learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114518758B (en) |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2014211667A (en) * | 2013-04-17 | 2014-11-13 | 日本電信電話株式会社 | Robot cooperative conveyance planning device, method, and program |
CN105306176A (en) * | 2015-11-13 | 2016-02-03 | 南京邮电大学 | Realization method for Q learning based vehicle-mounted network media access control (MAC) protocol |
CN107610235A (en) * | 2017-08-21 | 2018-01-19 | 北京精密机电控制设备研究所 | A kind of mobile platform navigation method and apparatus based on deep learning |
CN109726866A (en) * | 2018-12-27 | 2019-05-07 | 浙江农林大学 | Unmanned boat paths planning method based on Q learning neural network |
CN110378439A (en) * | 2019-08-09 | 2019-10-25 | 重庆理工大学 | Single robot path planning method based on Q-Learning algorithm |
CN110673637A (en) * | 2019-10-08 | 2020-01-10 | 福建工程学院 | Unmanned aerial vehicle pseudo path planning method based on deep reinforcement learning |
CN111098852A (en) * | 2019-12-02 | 2020-05-05 | 北京交通大学 | Parking path planning method based on reinforcement learning |
CN112344944A (en) * | 2020-11-24 | 2021-02-09 | 湖北汽车工业学院 | Reinforced learning path planning method introducing artificial potential field |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11106211B2 (en) * | 2018-04-02 | 2021-08-31 | Sony Group Corporation | Vision-based sample-efficient reinforcement learning framework for autonomous driving |
-
2022
- 2022-02-08 CN CN202210118037.9A patent/CN114518758B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2014211667A (en) * | 2013-04-17 | 2014-11-13 | 日本電信電話株式会社 | Robot cooperative conveyance planning device, method, and program |
CN105306176A (en) * | 2015-11-13 | 2016-02-03 | 南京邮电大学 | Realization method for Q learning based vehicle-mounted network media access control (MAC) protocol |
CN107610235A (en) * | 2017-08-21 | 2018-01-19 | 北京精密机电控制设备研究所 | A kind of mobile platform navigation method and apparatus based on deep learning |
CN109726866A (en) * | 2018-12-27 | 2019-05-07 | 浙江农林大学 | Unmanned boat paths planning method based on Q learning neural network |
CN110378439A (en) * | 2019-08-09 | 2019-10-25 | 重庆理工大学 | Single robot path planning method based on Q-Learning algorithm |
CN110673637A (en) * | 2019-10-08 | 2020-01-10 | 福建工程学院 | Unmanned aerial vehicle pseudo path planning method based on deep reinforcement learning |
CN111098852A (en) * | 2019-12-02 | 2020-05-05 | 北京交通大学 | Parking path planning method based on reinforcement learning |
CN112344944A (en) * | 2020-11-24 | 2021-02-09 | 湖北汽车工业学院 | Reinforced learning path planning method introducing artificial potential field |
Non-Patent Citations (2)
Title |
---|
张宁.基于强化Q学习和BP神经网络的移动机器人局部路径规划行为研究.《中国优秀硕士学位论文全文数据库 信息科技辑》.2021,正文第1-62页. * |
段建民.利用先验知识的Q-Learning路径规划算法研究.《电光与控制》.2019,第26卷(第9期),第29-33页. * |
Also Published As
Publication number | Publication date |
---|---|
CN114518758A (en) | 2022-05-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110320809B (en) | AGV track correction method based on model predictive control | |
JP5429901B2 (en) | Robot and information processing apparatus program | |
CN113110457B (en) | Autonomous coverage inspection method for intelligent robot in indoor complex dynamic environment | |
KR20210116576A (en) | How to use a visual robot-based map of the past | |
JP5068840B2 (en) | Robot program and information processing apparatus program | |
CN113296520B (en) | Routing inspection robot path planning method integrating A and improved gray wolf algorithm | |
CN110207706A (en) | A kind of automatic homing chair path planning algorithm based on grating map | |
CN116382304B (en) | DQN model-based multi-inspection robot collaborative path planning method and system | |
CN114595898A (en) | Intelligent ship control method considering channel environment | |
CN113062601A (en) | Q learning-based concrete distributing robot trajectory planning method | |
CN109352655A (en) | A kind of deformation-compensated method of robot returned based on multi output Gaussian process | |
CN112558601A (en) | Robot real-time scheduling method and system based on Q-learning algorithm and water drop algorithm | |
CN114518758B (en) | Indoor measurement robot multi-target point moving path planning method based on Q learning | |
CN109033181B (en) | Wind field geographic numerical simulation method for complex terrain area | |
CN116661469B (en) | Robot track error correction method and system | |
CN116576868A (en) | Multi-sensor fusion accurate positioning and autonomous navigation method | |
CN113910221B (en) | Mechanical arm autonomous motion planning method, device, equipment and storage medium | |
CN116817909A (en) | Unmanned aerial vehicle relay type navigation method based on deep reinforcement learning | |
CN115983560A (en) | Self-optimization collaborative driving control brewing workshop intelligent hoisting system | |
CN115790600A (en) | Algorithm of task coverage strategy for long-term inspection of robot in large range based on digital twin | |
CN113885494A (en) | Path planning method for electric power system robot | |
CN114264298A (en) | Navigation path generation method, device and medium for indoor map | |
CN111323036B (en) | Method and system for intelligently optimizing path of stock yard, electronic equipment and storage medium | |
CN117494919B (en) | Path planning method and device based on multi-robot collaborative stacking operation | |
CN105116724A (en) | Neural network optimization linear motor positioning force identification method based on genetic algorithm |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |