CN113062601B - Q learning-based concrete distributing robot trajectory planning method - Google Patents
Q learning-based concrete distributing robot trajectory planning method Download PDFInfo
- Publication number
- CN113062601B CN113062601B CN202110284547.9A CN202110284547A CN113062601B CN 113062601 B CN113062601 B CN 113062601B CN 202110284547 A CN202110284547 A CN 202110284547A CN 113062601 B CN113062601 B CN 113062601B
- Authority
- CN
- China
- Prior art keywords
- robot
- path
- matrix
- pouring
- action
- 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
Images
Classifications
-
- E—FIXED CONSTRUCTIONS
- E04—BUILDING
- E04G—SCAFFOLDING; FORMS; SHUTTERING; BUILDING IMPLEMENTS OR AIDS, OR THEIR USE; HANDLING BUILDING MATERIALS ON THE SITE; REPAIRING, BREAKING-UP OR OTHER WORK ON EXISTING BUILDINGS
- E04G21/00—Preparing, conveying, or working-up building materials or building elements in situ; Other devices or measures for constructional work
- E04G21/02—Conveying or working-up concrete or similar masses able to be heaped or cast
- E04G21/04—Devices for both conveying and distributing
- E04G21/0418—Devices for both conveying and distributing with distribution hose
- E04G21/0445—Devices for both conveying and distributing with distribution hose with booms
- E04G21/0463—Devices for both conveying and distributing with distribution hose with booms with boom control mechanisms, e.g. to automate concrete distribution
Landscapes
- Engineering & Computer Science (AREA)
- Architecture (AREA)
- Mechanical Engineering (AREA)
- Civil Engineering (AREA)
- Structural Engineering (AREA)
- Manipulator (AREA)
- Numerical Control (AREA)
Abstract
The invention relates to a novel track planning scheme of an intelligent concrete distributing robot, which is suitable for the autonomous pouring control of the concrete distributing robot, avoids the complex inverse kinematics interpolation calculation and belongs to the field of intelligent manufacturing. The invention designs a universal track planning frame, which comprises a rapid movement process that a material distribution robot is reset from an initial state to a path starting point and from a path end point to the initial state; and the material distributing robot performs a continuous concrete pouring process from the pouring path starting point to the pouring path end point. In the process of rapid movement, a simple interior point method is adopted to perform inverse solution optimization with time optimization as a target, and a cubic polynomial is adopted to fit a track. In the continuous concrete pouring process, an error band with a certain area is formed on a path to be poured at the tail end of the distributing robot, the formed path error band is divided into regions by using a Q learning algorithm, reward values are given to the divided regions according to pouring targets and constraints, Q value training is carried out on a given grid, finally, action sequences of all joints of the robot are formed, the robot action is directly obtained, and the complex track planning process based on kinematic inverse solution optimization is avoided.
Description
Technical Field
The invention relates to a novel track planning method of an intelligent concrete distributing robot, which is suitable for the autonomous pouring control of the concrete distributing robot, avoids complex interpolation calculation and belongs to the field of intelligent construction.
Background
The concrete distributing robot is a building industrial robot for conveying concrete to a construction site, and plays a very important role in the development process of urban modern construction. Along with the requirement of engineering construction on the efficiency of the material distribution robot is continuously improved, the intelligent control research on the material distribution robot is gradually developed. The intelligent control is inseparable from the path and track planning of the mechanical arm, for an industrial robot, the path planning mainly refers to the track of the tail end motion of the mechanical arm, and the track planning refers to curves of displacement, speed, acceleration and the like when the operated arm frames perform combined motion. The specific work of the material distribution robot is to realize the displacement of the pumping port at the tail end of the arm support, and the displacement can be analyzed according to the mechanism of the industrial robot. The cloth pouring surface is generally divided into a non-rotary horizontal plane, a vertical cylindrical surface and a non-rotary spatial plane or curved surface, the target pouring routes of the cloth are different according to different pouring surfaces, and in order to simplify the pouring route planning, the general routes are composed of straight lines.
In recent years, intelligent real-time dynamic autonomous planning plays an extremely important role in operations of industrial robots and the like. Aiming at the problem of planning of concrete distribution, the intelligent real-time dynamic path planning of the automatic concrete distribution robot is combined, and the pouring efficiency and quality of the distribution task can be further improved. At present, for a large-scale distribution boom system, the construction condition is relatively severe, the calculation amount of the traditional track planning method is large, and the optimal performance index is difficult to determine, so that the traditional track planning method cannot cope with real-time variation factors of a construction site, in the face of construction surfaces with different shapes, the pouring path of a working section is often planned and judged according to the limited visual angle of site workers, the quality of construction pouring excessively depends on the operation experience and the technical level of workers, the automation degree is low, and the track movement requirement of the pouring end of the large-scale distribution boom system cannot be met. The reinforcement learning is used as a branch of a machine learning method, the basic principle of the reinforcement learning is to simulate the learning process of an organism, and an intelligent agent obtains learning experience in continuous interaction with the external environment to gradually train the autonomous planning capability of the intelligent agent. Technically, the reinforcement learning method is similar to the learning operation process of workers, and is an effective way for meeting the autonomous path planning requirement of the work end of the material distribution robot.
Disclosure of Invention
The invention designs a track planner based on Q learning aiming at the autonomous planning and control precision requirements of an intelligent redundant material distribution robot. The method comprises the following steps that a path required by a material distribution robot to be subjected to trajectory planning is a known condition, and firstly, a joint trajectory is planned by a material distribution machine from an initial position to a path starting point by adopting an inner point method with optimal time; when the track planning is carried out on the known straight path, an error band with the bandwidth of 2 times of precision is established by taking the known pouring path as the center according to the pouring precision requirement, the error band is divided in a gridding mode based on a Q learning algorithm in reinforcement learning, the divided grid is given an antipodal scalar reward value according to the working condition of the cloth robot and general indexes, the reward value is the maximum target, distribution training is carried out, the optimal observation-action-reward sequence in the error band environment is obtained, and the joint planning track of the cloth robot is formed according to the action sequence.
The intelligent concrete distributing robot has redundancy, and the robot is a continuous casting straight line or curve path when distributing, so the path planning scheme for designing the distributing robot is carried out in a Cartesian space, and the planning problems of continuous path casting and redundancy freedom degree are considered. According to the existing planning method, firstly, a DH coordinate system transformation method is adopted to establish a positive kinematics model for the robot, on the premise of a known pouring path in a Cartesian space, the path is firstly dispersed according to a certain quantity value, namely, the path is divided into a plurality of sections of paths, the starting point and the end point of each discrete section of path are subjected to kinematics inverse solution operation, namely, the robot kinematics inverse derivation is carried out from the pouring position at the tail end, and the change angle of each joint of the mechanical arm is obtained. Due to the existence of redundant degrees of freedom, the number of the transformable joints of the robot is larger than that of the degrees of freedom in the motion space, the inverse solution operation of kinematics can generate a multi-solution condition, generally, a target optimization algorithm is adopted to select the optimal inverse solution, and the optimization target is set to be optimal in time or energy. After the inverse kinematics solution operation is completed, a series of joint angle values are obtained, data fitting is carried out on the joint angle values, the fitting method uses cubic polynomial fitting, quintic polynomial fitting or B spline curve fitting for multiple purposes, the movement track of each joint of the mechanical arm is obtained, and the track planning task is completed.
The conventional method has many problems for a robot arm which has redundancy and performs continuous linear motion. Firstly, after the trajectory planning is completed, positive kinematic operation is carried out on the planned trajectory again, namely a path in a Cartesian space is deduced from the space trajectory of each joint, and the result shows that although discrete points of the path can be reached, the deviation between the middle section of the discrete points and the actual path is large, particularly in an overlong large mechanical arm such as a distributing robot, the trajectory error of the traditional method can reach about 1m, so that the traditional concrete distributing robot is poor in autonomy and difficult to achieve the goal of intelligent manufacturing due to the fact that the traditional concrete distributing robot is used for manually operating a pouring tail end to control the actual path range. For the situation, the existing research focuses on the fitting of a higher-order polynomial or the research of performing path point segmentation fitting, such as a three-five-three-order polynomial interpolation method, which needs higher mathematical skill and has high computational difficulty, and is difficult to implement in the real-time planning of an industrial robot. Due to the existence of redundancy, the inverse kinematics solution is solved by adopting an optimization algorithm, the optimization algorithm mostly aims at time optimization or energy optimization, weight and constraint can be added in the design of a target function for coupling, so that the multi-objective optimization requirement is met, the algorithm is complex and large in calculation amount, and greatly depends on the correct setting of performance indexes, the working condition of the distributing robot is complex, the performance indexes and the constraint conditions change along with the change of the construction environment at any time, so that the algorithm design is difficult, the internal structure of the complex algorithm is difficult to change, and a fixed algorithm frame with generalization cannot be found.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and discloses a concrete distributing robot track planning method based on Q learning.
The technical scheme adopted by the invention for solving the technical problem is as follows:
aiming at the characteristics of a redundant concrete distribution robot, a universal track planning frame is designed, and the track planning of the distribution robot is divided into two parts: one part is a rapid movement process of the cloth robot from an initial state to a path starting point and from a path end point to an initial state; and the other part is the process that the material distribution robot carries out continuous concrete pouring from the starting point of the pouring path to the end point of the pouring path. In the process of rapid movement, because a middle path does not need to be considered, a simple interior point method is adopted to perform inverse solution optimization with time optimization as a target, and a cubic polynomial is adopted to fit a track. In the continuous pouring process of concrete, an error band with a certain area is formed on a path to be poured at the tail end of the distributing robot, and the error band width is set according to a given pouring precision condition. And performing region division on the formed path error band by using a Q learning algorithm, giving reward values to the divided regions according to a poured target and constraint, training a given grid, finally forming an action sequence of each joint of the robot, directly obtaining the action of the robot, and avoiding a complex track planning process.
The invention has the beneficial effects that: aiming at the working characteristics of the cloth robot, the trajectory planner is designed, the universality and the generalization are better, the internal parameters are mainly set scalar reward values, and the change and the test are easy. The track planning is carried out by adopting a mode of designing error bands for pouring paths, the error size can be set independently according to the working precision requirement, the actual path deviation is ensured to be within the working precision requirement range, and the problem of overlarge errors among path points in the traditional interpolation planning method is solved. The motion values of all joints of the robot can be directly obtained by adopting a Q learning algorithm for training, so that the complex multi-target optimization inverse solution operation and data fitting process is avoided. The on-line learning mode is adopted for planning, so that the autonomy of the concrete distributing machine is improved, and the aim of unmanned engineering machinery in intelligent construction is easily achieved.
Drawings
FIG. 1 is an overall structure diagram of an intelligent redundant concrete distribution robot (a three-joint rotary concrete distribution robot, previously filed patent 2020111625562);
FIG. 2 is a diagram of a concrete placement robot trajectory planner based on Q-learning;
FIG. 3 is a flow chart of trajectory planning for a continuous straight pouring process;
FIG. 4 is an exemplary diagram of a straight-line trajectory planning using a trajectory planner.
Detailed Description
Fig. 1 shows the overall structure of a designed intelligent redundant concrete distribution robot, which mainly comprises five modules, including a column assembly (1), a pipeline assembly (2), a pipe clamp (3), a cantilever support (4) and a turntable assembly (5). The design adopts three rotary joints, namely three turntable assemblies (5), and one degree of freedom is redundant in the design aiming at two-degree-of-freedom plane pouring.
Fig. 2 is a general structure of the trajectory planner, including a fast moving part and a summary of the planning method of the concrete continuous casting part.
Fig. 3 is a detailed description of an idea layer and a technical layer of a track planning section in a concrete continuous casting process, in which a Q learning method is adopted, and fig. 3 is a block diagram and a summary of an application manner and a process thereof.
In fig. 2, a concrete distribution robot trajectory planning overall design framework based on Q learning is shown, and the design framework is divided according to layers and mainly comprises a structural layer, an idea layer, a technical layer and an output layer. In a structural layer, a concrete distributing robot trajectory planner based on Q learning is divided into two parts: one part is a rapid movement process of the cloth robot from an initial state to a path starting point and from a path end point to an initial state; and the other part is the process that the material distribution robot carries out continuous concrete pouring from the starting point of the pouring path to the end point of the pouring path.
Based on the division of the structural layer, the following ideas are adopted: in the process of rapid movement, because an intermediate path does not need to be considered in a general situation, a traditional inverse kinematics trajectory planning idea is adopted to perform inverse solution optimization with time optimization as a target, a simple interior point method is adopted in the technology, a cubic polynomial fitting trajectory can be adopted subsequently, and a trajectory curve of each joint angle is finally generated; in the concrete continuous pouring process, the idea that an error band with a certain area is formed on a path to be poured at the tail end of a distributing robot is planned, the error bandwidth is set according to a given pouring precision condition and is mostly 2-time precision, a Q learning algorithm is adopted in the technology, the formed path error band is divided into areas, the divided areas are given reward values according to pouring targets and constraints, given lattices are trained, action sequences of joints of the robot are finally generated, robot actions are directly obtained, and the complex track planning process is avoided.
The continuous concrete pouring process comprises two processes relative to the rapid movement process, wherein the first process is the rapid movement process, and the second process is the continuous concrete pouring process. The fast moving process means that the initial position is not necessarily at the moving starting point of the set pouring track when the material distribution robot starts to work, so that the fast moving process is set to enable the initial position of the material distribution robot to return to the starting point of the pouring track.
In fig. 3, a path planning process of the material distributing robot in the continuous straight pouring process of the concrete is shown.
Firstly, planning a path according to a construction environment, generally setting a pouring track as a straight line, then determining a construction precision requirement, and establishing a track planning error band which takes the path as a regional center line and takes 2 times of precision as a regional width. Considering that in the path optimization process, the shortest path does not need to be found, but a suboptimal solution is found under the condition of balancing efficiency and route quality, an error band is established to sacrifice part of precision, the efficiency and the route quality are balanced under the condition of a certain width of the error band, and the error is controllable, so that the requirement of track planning can be met.
And secondly, dividing grids after determining the error band, establishing a dynamic reward value model R according to a grid region, wherein the reward value of each grid is stored in a matrix form, setting the pouring tail end of the material distribution robot to be positive reward towards a path target terminal point, otherwise setting the pouring tail end of the material distribution robot to be negative reward, and setting the reward of regions outside the error band to be negative infinity so as to ensure that the material distribution robot always moves in the error band region in the planning process.
And thirdly, quantizing the reward values in the R matrix so as to facilitate the subsequent planner to learn.
3.1 setting that the robot moves once, subtracting one unit from the reward value, and ensuring the optimal energy requirement of the robot, namely achieving the casting track requirement with the least movement.
And 3.2 adding scalar rewards deviating from the central value of the path and scalar rewards of finished action distances into the R matrix, namely storing the track error of the distributing robot in each state in the R matrix, directly taking an error value as an equivalent scalar negative reward, and taking the distance between the pouring position of the robot in each state and the starting point of the path as an equivalent scalar positive reward to ensure that the distributing robot moves to a target point in an error range.
3.3 setting the grid where the path terminal point is located as the maximum reward value, and ensuring that the planner always moves forward towards the target terminal point during the tracing planning.
And fourthly, establishing an initial R matrix according to the requirements, and establishing a motion matrix a of the cloth robot, wherein the motion comprises 3 types of forward motion, static motion and backward motion, so that the specification of the motion matrix is 3 multiplied by 3, the motion matrix represents 27 states of the cloth robot under all feasible motions, and in order to prevent a dead zone of a long-term static state, a (0,0,0) state is removed, namely all joint static states are removed, namely the motion matrix a comprises 26 feasible motion states of the cloth robot.
And fifthly, establishing a dynamic Q matrix according to the specified reward matrix R and the action matrix a, wherein the Q matrix is equivalent to an action-value function, inputting the current state of the robot and the next action to be performed to obtain all reward values reaching a target point, and initializing the reward values to be 0 in all grids, and setting the grid Q value of the post-marking terminal point to be 100 units.
According to the steps, a Q learning updating strategy formula Q (s, a) ═ E [ R + gamma maxQ (s ', a') | s, a ] dynamic Q matrix is adopted to continuously train and iterate, wherein:
scalar rewards deviating from the central value of the path and scalar rewards of finished action distances are added into the R matrix, namely, the track error of the distributing robot in each state is stored in the R matrix, the error value is directly used as the equivalent scalar negative reward, the distance between the pouring position of the robot in each state and the starting point of the path is used as the equivalent scalar positive reward, and the distributing robot is ensured to move to the target point in the error range;
the action state s comprises three rotation joint angles, namely three angles of the turntable assembly (5);
the motion matrix a comprises 3 types of forward motion, static motion and backward motion, the specification of the motion matrix is 3 multiplied by 3, the motion matrix represents 27 states of the cloth robot under all feasible motions, in order to prevent a long-term static state dead zone, a (0,0,0) state is removed, namely all joint static states are avoided, namely the motion matrix a comprises 26 feasible motion states s of the cloth robot;
expectation in mathematical probability E;
q (s ', a') is the maximum Q value obtained in one training, gamma is a learning rate, the value is 0.9, the updating is stopped until the set iteration times reach the upper limit or the Q matrix is converged, the obtained Q matrix is an optimal matrix by default, and a series of optimal (action value and reward value) mapping relations can be obtained according to the optimal matrix, so that 3 joint action value sequences of the distributing robot obtained by path planning in a set error zone are obtained, the path planning of straight line casting is completed, the complex inverse solution optimization operation of kinematics is avoided, and the optimal path point and the state-action sequence of the distributing robot of the target casting path are obtained.
As shown in fig. 4, the optimal continuous path planning points are obtained by the Q learning trajectory planner of the present application, and the optimal continuous trajectory can be obtained by fitting the portions between the path points in a conventional curve fitting manner. The concrete continuous pouring track obtained by the method provided by the application is composed of a series of discrete points, and each discrete point represents a motion state of the material distribution robot.
In fig. 4, all path points of a straight path obtained by the concrete distributing robot trajectory planner based on Q learning are all states reached by the distributing robot and are within a set error band range, so that the requirements of autonomous planning and error control of the intelligent construction engineering machinery are met.
Claims (1)
1. The utility model provides a concrete cloth robot path planning method based on Q study, to the characteristics of redundancy concrete cloth robot, its characterized in that designs a general path planning frame, divides cloth robot path planning into two parts: one part is a rapid movement process of the cloth robot from an initial state to a path starting point and from a path end point to an initial state; the other part is the process that the material distribution robot carries out continuous concrete pouring from the starting point of the pouring path to the end point of the pouring path;
the design adopts three rotary joints, namely three turntable assemblies (5);
the track planning process of the material distribution robot in the concrete continuous straight pouring process comprises the following steps:
firstly, planning a path according to a construction environment, and establishing a track planning error band which takes the path as a regional center line and takes 2 times of precision as the regional width;
secondly, dividing grids after the error band is determined, establishing a dynamic reward value model R according to grid areas, wherein the reward value of each grid is stored in a matrix form, setting the pouring tail end of the cloth robot to face a path target terminal point as a positive reward, otherwise setting the pouring tail end of the cloth robot as a negative reward, and setting the reward of areas outside the error band as a negative infinity;
thirdly, quantizing the reward values in the R matrix;
3.1, setting that the reward value is subtracted by one unit every time the robot acts once, and ensuring the optimal energy requirement of the robot, namely, the pouring track requirement is met by the least actions;
3.2 adding scalar rewards deviating from the central value of the path and scalar rewards of finished action distances into the R matrix, namely storing the track error of the distributing robot in each state in the R matrix, directly taking an error value as an equivalent scalar negative reward, and taking the distance between a pouring position of the robot in each state and the starting point of the path as an equivalent scalar positive reward to ensure that the distributing robot moves to a target point in an error range;
3.3 setting the grid where the path end point is located as the maximum reward value;
fourthly, establishing an initial R matrix according to the requirements, and establishing a cloth robot action matrix a, wherein the action comprises 3 types of forward movement, static movement and backward movement, so that the specification of the action matrix is 3 multiplied by 3, the action matrix represents 27 states of the cloth robot under all feasible actions, and in order to prevent a long-term static state dead zone, the (0,0,0) state is removed, namely all joint static states are removed, namely the action matrix a comprises 26 feasible action states s of the cloth robot;
fifthly, establishing a dynamic Q matrix according to a specified reward matrix R and an action matrix a, wherein the Q matrix is equivalent to an action-value function, and inputting the current state of the robot and the next action to be performed to obtain all reward values reaching a target point;
according to the steps, adopting a Q reinforcement learning updating strategy formula Q (s, a) ═ E [ R + gamma maxQ (s ', a') | s, a ] dynamic Q matrix to train and iterate continuously, wherein:
scalar rewards deviating from the central value of the path and scalar rewards of finished action distances are added into the R matrix, namely, the track error of the distributing robot in each state is stored in the R matrix, the error value is directly used as the equivalent scalar negative reward, the distance between the pouring position of the robot in each state and the starting point of the path is used as the equivalent scalar positive reward, and the distributing robot is ensured to move to the target point in the error range;
the action state s comprises three rotation joint angles, namely three angles of the turntable assembly (5);
the motion matrix a comprises 3 types of forward motion, static motion and backward motion, the specification of the motion matrix is 3 multiplied by 3, the motion matrix represents 27 states of the cloth robot under all feasible motions, in order to prevent a long-term static state dead zone, a (0,0,0) state is removed, namely all joint static states are avoided, namely the motion matrix a comprises 26 feasible motion states s of the cloth robot;
expectation in mathematical probability E;
q (s ', a') is the maximum Q value obtained in one training, gamma is a learning rate, the value is 0.9, the updating is stopped until the set iteration times reach the upper limit or the Q matrix is converged, the obtained Q matrix is an optimal matrix by default, and a series of optimal (action value and reward value) mapping relations can be obtained according to the optimal matrix, so that 3 joint action value sequences of the distributing robot obtained by path planning in a set error zone are obtained, the path planning of straight line casting is completed, the complex inverse solution optimization operation of kinematics is avoided, and the optimal path point and the state-action sequence of the distributing robot of the target casting path are obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110284547.9A CN113062601B (en) | 2021-03-17 | 2021-03-17 | Q learning-based concrete distributing robot trajectory planning method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110284547.9A CN113062601B (en) | 2021-03-17 | 2021-03-17 | Q learning-based concrete distributing robot trajectory planning method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113062601A CN113062601A (en) | 2021-07-02 |
| CN113062601B true CN113062601B (en) | 2022-05-13 |
Family
ID=76560916
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110284547.9A Active CN113062601B (en) | 2021-03-17 | 2021-03-17 | Q learning-based concrete distributing robot trajectory planning method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113062601B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113377007A (en) * | 2021-06-17 | 2021-09-10 | 同济大学 | Fuzzy neural network-based concrete distributing robot control method |
| CN114351541B (en) * | 2021-09-27 | 2023-11-07 | 广西北投交通养护科技集团有限公司 | Automatic pouring system for foam lightweight concrete |
| CN113774744B (en) * | 2021-09-27 | 2022-11-11 | 广西北投交通养护科技集团有限公司 | Automatic pouring method and device for foam lightweight concrete |
| CN114021229B (en) * | 2021-10-26 | 2024-08-27 | 同济大学 | Path planning method, system and storage medium for bridge concrete distribution |
| CN114396160B (en) * | 2022-01-21 | 2024-03-15 | 武汉博雅宏科技有限公司 | Concrete spreader and control method thereof |
Citations (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102799179A (en) * | 2012-07-06 | 2012-11-28 | 山东大学 | Mobile robot path planning algorithm based on single-chain sequential backtracking Q-learning |
| CN103792956A (en) * | 2014-01-26 | 2014-05-14 | 三一汽车制造有限公司 | Method and device controlling spatial motion of intelligent arm |
| US8972057B1 (en) * | 2013-01-09 | 2015-03-03 | The Boeing Company | Systems and methods for generating a robotic path plan in a confined configuration space |
| CN105234930A (en) * | 2015-10-15 | 2016-01-13 | 哈尔滨工程大学 | Control method of motion of redundant mechanical arm based on configuration plane |
| CN108089441A (en) * | 2017-12-08 | 2018-05-29 | 中国科学院长春光学精密机械与物理研究所 | Clap machine secondary mirror six degree of freedom accurate adjusting mechanism calibration algorithm and storage medium in space |
| CN108297105A (en) * | 2018-01-17 | 2018-07-20 | 广东工业大学 | A kind of mechanical arm task level time optimal trajectory planning method |
| CN108873687A (en) * | 2018-07-11 | 2018-11-23 | 哈尔滨工程大学 | A kind of Intelligent Underwater Robot behavior system knot planing method based on depth Q study |
| CN109434841A (en) * | 2019-01-07 | 2019-03-08 | 南京航空航天大学 | A kind of length of a game's optimal trajectory planning method of industrial robot dynamic speed adjustment |
| CN109478057A (en) * | 2016-05-09 | 2019-03-15 | 强力物联网投资组合2016有限公司 | Method and system for industrial Internet of Things |
| CN109910015A (en) * | 2019-04-12 | 2019-06-21 | 成都天富若博特科技有限责任公司 | The terminal end path planning algorithm of construction robot is smeared in a kind of mortar spray |
| CN109944448A (en) * | 2019-04-19 | 2019-06-28 | 长沙赛搏机器智能有限公司 | A kind of concrete pouring robot, distributing method and concrete pumping equipment |
| CN110032189A (en) * | 2019-04-22 | 2019-07-19 | 河海大学常州校区 | A kind of intelligent storage method for planning path for mobile robot not depending on map |
| WO2019165451A1 (en) * | 2018-02-26 | 2019-08-29 | Nvidia Corporation | Systems and methods for computer-assisted shuttles, buses, robo-taxis, ride-sharing and on-demand vehicles with situational awareness |
| CN110378439A (en) * | 2019-08-09 | 2019-10-25 | 重庆理工大学 | Single robot path planning method based on Q-Learning algorithm |
| CN110732601A (en) * | 2019-09-27 | 2020-01-31 | 上海应用技术大学 | Manipulator-assisted feeding and discharging and obstacle avoidance detection system and method based on visual detection |
| CN111065494A (en) * | 2016-07-15 | 2020-04-24 | 快砖知识产权私人有限公司 | Robot base path planning |
| CN111891928A (en) * | 2020-06-09 | 2020-11-06 | 东南大学 | Architecture of unmanned intelligent hoisting system |
| CN111983924A (en) * | 2020-07-29 | 2020-11-24 | 华南理工大学 | Robot time optimal trajectory planning method and controller based on reinforcement learning |
| CN112000009A (en) * | 2020-07-27 | 2020-11-27 | 南京理工大学 | Material transfer device reinforcement learning control method based on state and disturbance estimation |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102525795B (en) * | 2012-01-16 | 2014-07-09 | 沈阳理工大学 | Fast automatic positioning method of foot massaging robot |
| CN104932264B (en) * | 2015-06-03 | 2018-07-20 | 华南理工大学 | The apery robot stabilized control method of Q learning frameworks based on RBF networks |
| US20190184561A1 (en) * | 2017-12-15 | 2019-06-20 | The Regents Of The University Of California | Machine Learning based Fixed-Time Optimal Path Generation |
-
2021
- 2021-03-17 CN CN202110284547.9A patent/CN113062601B/en active Active
Patent Citations (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102799179A (en) * | 2012-07-06 | 2012-11-28 | 山东大学 | Mobile robot path planning algorithm based on single-chain sequential backtracking Q-learning |
| US8972057B1 (en) * | 2013-01-09 | 2015-03-03 | The Boeing Company | Systems and methods for generating a robotic path plan in a confined configuration space |
| CN103792956A (en) * | 2014-01-26 | 2014-05-14 | 三一汽车制造有限公司 | Method and device controlling spatial motion of intelligent arm |
| CN105234930A (en) * | 2015-10-15 | 2016-01-13 | 哈尔滨工程大学 | Control method of motion of redundant mechanical arm based on configuration plane |
| CN109478057A (en) * | 2016-05-09 | 2019-03-15 | 强力物联网投资组合2016有限公司 | Method and system for industrial Internet of Things |
| CN111065494A (en) * | 2016-07-15 | 2020-04-24 | 快砖知识产权私人有限公司 | Robot base path planning |
| CN108089441A (en) * | 2017-12-08 | 2018-05-29 | 中国科学院长春光学精密机械与物理研究所 | Clap machine secondary mirror six degree of freedom accurate adjusting mechanism calibration algorithm and storage medium in space |
| CN108297105A (en) * | 2018-01-17 | 2018-07-20 | 广东工业大学 | A kind of mechanical arm task level time optimal trajectory planning method |
| WO2019165451A1 (en) * | 2018-02-26 | 2019-08-29 | Nvidia Corporation | Systems and methods for computer-assisted shuttles, buses, robo-taxis, ride-sharing and on-demand vehicles with situational awareness |
| CN108873687A (en) * | 2018-07-11 | 2018-11-23 | 哈尔滨工程大学 | A kind of Intelligent Underwater Robot behavior system knot planing method based on depth Q study |
| CN109434841A (en) * | 2019-01-07 | 2019-03-08 | 南京航空航天大学 | A kind of length of a game's optimal trajectory planning method of industrial robot dynamic speed adjustment |
| CN109910015A (en) * | 2019-04-12 | 2019-06-21 | 成都天富若博特科技有限责任公司 | The terminal end path planning algorithm of construction robot is smeared in a kind of mortar spray |
| CN109944448A (en) * | 2019-04-19 | 2019-06-28 | 长沙赛搏机器智能有限公司 | A kind of concrete pouring robot, distributing method and concrete pumping equipment |
| CN110032189A (en) * | 2019-04-22 | 2019-07-19 | 河海大学常州校区 | A kind of intelligent storage method for planning path for mobile robot not depending on map |
| CN110378439A (en) * | 2019-08-09 | 2019-10-25 | 重庆理工大学 | Single robot path planning method based on Q-Learning algorithm |
| CN110732601A (en) * | 2019-09-27 | 2020-01-31 | 上海应用技术大学 | Manipulator-assisted feeding and discharging and obstacle avoidance detection system and method based on visual detection |
| CN111891928A (en) * | 2020-06-09 | 2020-11-06 | 东南大学 | Architecture of unmanned intelligent hoisting system |
| CN112000009A (en) * | 2020-07-27 | 2020-11-27 | 南京理工大学 | Material transfer device reinforcement learning control method based on state and disturbance estimation |
| CN111983924A (en) * | 2020-07-29 | 2020-11-24 | 华南理工大学 | Robot time optimal trajectory planning method and controller based on reinforcement learning |
Non-Patent Citations (1)
| Title |
|---|
| 基于区域扩张策略的势场强化学习算法路径规划研究;阳杰等;《微处理机》;20210228;第42卷(第1期);第47-51页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113062601A (en) | 2021-07-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113062601B (en) | Q learning-based concrete distributing robot trajectory planning method | |
| CN104062902B (en) | Delta robot time optimal trajectory planning method | |
| CN105511266B (en) | The Delta method for planning track of robot of particle cluster algorithm is searched for based on gravitation | |
| CN103692433B (en) | Model decoupling three-arm-lever five-freedom-degree translation welding robot and decoupling method thereof | |
| CN102646148B (en) | Motion trajectory planning method of mechanical arm of humanoid robot for preventing collision | |
| CN110815180A (en) | Six-degree-of-freedom parallel robot motion analysis modeling and fast solving method | |
| CN106647282A (en) | Six-freedom-degree robot track planning method giving consideration to tail end motion error | |
| CN109397292B (en) | Analytical solution-based 7-degree-of-freedom mechanical arm control method and system | |
| CN111610523B (en) | Parameter correction method for wheeled mobile robot | |
| CN108326849A (en) | A kind of multi-degree-of-freemechanical mechanical arm dynamic obstacle avoidance paths planning method based on modified embedded-atom method | |
| CN110455290B (en) | Optimal trajectory planning method for intelligent hydraulic excavator | |
| CN112487569A (en) | Solving method for working space of mobile robot capable of reaching fixed time length | |
| CN104570735A (en) | Arc-shaped movement-track algorithm for palletizing robot | |
| CN105773602A (en) | Control system for palletizing robot | |
| CN116652936A (en) | Continuous casting ladle mechanical arm track multi-objective optimization method based on snake optimization algorithm | |
| CN113848725A (en) | Mechanical arm time optimal trajectory planning and optimizing method based on adaptive genetic algorithm | |
| CN113467475B (en) | Mecanum wheel omnidirectional mobile robot track tracking sliding mode control method | |
| CN117021076A (en) | Large redundant mechanical arm track planning method considering energy consumption and motion performance | |
| CN112549008B (en) | Projection kinematics control method of large heavy-load mechanical arm | |
| CN117415806A (en) | Obstacle avoidance track planning method for mechanical arm for loading and unloading drill rod under variable target condition | |
| CN109278041B (en) | Transition error controllable mechanical arm linear track self-adaptive linking planning method | |
| CN111546379A (en) | Sensor layout method for monitoring operating state of industrial robot | |
| CN114896553B (en) | Forward kinematics geometric solving method for Delta type parallel robot | |
| CN113829351B (en) | Cooperative control method of mobile mechanical arm based on reinforcement learning | |
| CN114986514A (en) | Parallel mechanism track planning method considering joint friction |
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 |