CN114995397A - ROS (reactive oxygen species) platform-oriented multi-robot formation algorithm based on distributed asynchronous optimization - Google Patents
ROS (reactive oxygen species) platform-oriented multi-robot formation algorithm based on distributed asynchronous optimization Download PDFInfo
- Publication number
- CN114995397A CN114995397A CN202210520624.0A CN202210520624A CN114995397A CN 114995397 A CN114995397 A CN 114995397A CN 202210520624 A CN202210520624 A CN 202210520624A CN 114995397 A CN114995397 A CN 114995397A
- Authority
- CN
- China
- Prior art keywords
- robot
- ros
- optimization
- distributed
- condition
- 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.)
- Pending
Links
- 238000005457 optimization Methods 0.000 title claims abstract description 31
- 230000015572 biosynthetic process Effects 0.000 title claims abstract description 24
- 239000003642 reactive oxygen metabolite Substances 0.000 title 2
- 239000011159 matrix material Substances 0.000 claims description 7
- 238000004891 communication Methods 0.000 abstract description 12
- 230000001360 synchronised effect Effects 0.000 abstract description 9
- 230000000903 blocking effect Effects 0.000 abstract description 5
- 238000000034 method Methods 0.000 abstract description 4
- 238000012360 testing method Methods 0.000 description 5
- 230000007246 mechanism Effects 0.000 description 3
- 230000002776 aggregation Effects 0.000 description 2
- 238000004220 aggregation Methods 0.000 description 2
- 238000005034 decoration Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000001960 triggered effect Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
Images
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
- G05D1/0214—Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory in accordance with safety or protection criteria, e.g. avoiding hazardous areas
-
- 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
-
- 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
- G05D1/0287—Control of position or course in two dimensions specially adapted to land vehicles involving a plurality of land vehicles, e.g. fleet or convoy travelling
- G05D1/0291—Fleet control
-
- 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
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
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)
- Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)
Abstract
The invention discloses a ROS-oriented platform based distributed asynchronous optimization multi-robot formation algorithm, wherein each robot firstly determines a formation according to a target to be blocked and establishes a distributed optimization model; then, whether event triggering is met is judged by comparing whether the current updating progress of the robot is slower than that of other nodes, and if the condition is met, the robot summarizes the information of neighbors and updates the parameters of the robot; if the condition is not met, then the robot does not update and communicate; through the method, the node with the high updating rate can not gather neighbor node information for many times and further can not meet the constraint condition, the algorithm convergence is ensured to be correct on the premise of ROS-based asynchronous high-efficiency communication, the blocking behavior of the synchronous algorithm is avoided while the formation is kept, the algorithm convergence is faster, and the real-time performance of the robot system is higher.
Description
Technical Field
The invention belongs to the technical field of automation and computers, and particularly relates to a ROS platform-oriented multi-robot formation algorithm based on distributed asynchronous optimization.
Background
The cooperation/formation of multiple robots has very important practical significance, for example, in military/life cooperation, one object is enclosed, and the target object is moved and the formation is kept; in addition, for the aerial unmanned aerial vehicle, the target object can be shot in real time from multiple angles. Therefore, it is of great significance to design a dynamic queuing algorithm which can change according to the target to be enclosed and the self requirement.
The problem of optimization is not difficult to solve, but the traditional distributed optimization algorithm for solving the problem of constrained optimization is based on synchronous communication, is incompatible with the existing mainstream ROS platform based on the asynchronous communication robot system, and has many advantages.
Based on the fact that the communication mode of the published topic/the subscribed topic in the ROS platform is a natural asynchronous communication mode, the intelligent agent publishes own information through a publishing topic, the information before is covered once the published information is updated, the information of the neighbor is received through the subscribed topic, the pace is completely dependent on the intelligent agent, the received information is possibly advanced and delayed, the real-time performance is not realized, and the natural asynchronous environment is realized. However, if a traditional distributed optimization algorithm is deployed in the asynchronous environment, the time delay of information and the imbalance of updating destroy the designed constraint condition, and the error of the result of final convergence through experiments and the theoretical optimal solution can reach 50%, so that the existing optimization algorithm is not suitable for a platform based on ROS.
Disclosure of Invention
Aiming at the problem that the traditional algorithm in the prior art is not suitable for an ROS platform, the invention provides a distributed asynchronous optimization-based multi-robot formation algorithm for the ROS platform, wherein each robot firstly determines a formation according to a target to be blocked and establishes a distributed optimization model; then, whether event triggering is met is judged by comparing whether the current updating progress of the robot is slower than that of other nodes, and if the condition is met, the robot summarizes the information of neighbors and updates the parameters of the robot; if the condition is not met, then the robot does not update and communicate; through the method, the node with the high updating rate can not gather neighbor node information for many times and further can not meet the constraint condition, the algorithm convergence is ensured to be correct on the premise of ROS-based asynchronous high-efficiency communication, the blocking behavior of the synchronous algorithm is avoided while the formation is kept, the algorithm convergence is faster, and the real-time performance of the robot system is higher. In order to achieve the purpose, the invention adopts the technical scheme that: a ROS platform-oriented multi-robot formation algorithm based on distributed asynchronous optimization, which is applied to a system with i-1, 2, 3.
S1: determining a formation according to the position of a target to be enclosed and establishing a distributed optimization model, wherein the objective function of the distributed optimization model is the position x of the optimization robot i To the optimal observation position s i The distance of (d) is minimal, i.e.:
minf i (x i )=||x i -s i || 2
wherein x is i ∈R 2 Is the position coordinates of robot i; s i ∈R 2 Is the optimal observation position of the robot i;
the constraints of the distributed optimization model at least comprise: robot i needs to be in its safe area Ω i And observe the region S i The centers of the positions of the inner robot and all the robots are coincided with the target to be blocked, namely:
s.t x i ∈Ω i
x i ∈S i
wherein d ∈ R 2 Representing the position of the target to be enclosed;
s2: judging whether event triggering is met by comparing whether the current updating progress of the robot is slower than other nodes or not, if the updating times of the robot i are more than that of other nodes, stopping the robot i to wait for other agents; when the event trigger condition is satisfied, the robot i performs local update, and continues to step S3; the event triggering conditions of the robot i are as follows:
wherein k is i The number of local updates for robot i; k is a radical of formula j The number of local updates for robot j; τ is event triggerSpring parameters; n is a radical of i A neighbor set of a robot i;
s3: and local updating according to the updating condition of the robot i, wherein the specific formula is as follows:
k i =k i +1
where eta is the step length, a ij As a weight of the neighbor(s),is a gradient, v i Is the Lagrangian constant; i is 2 Is an identity matrix of 2 x 2,is kronecker product, A ═ a ij ] m*m A is an adjacency matrix of the robot, when i ≠ j ij >0 if and only if the robot i sends j information, i j,
s4: judging whether a termination condition is met, if so, terminating the algorithm and outputting a final position x i (ii) a If not, the step is returned to the step S3 to repeat the steps, and the termination condition is set as the upper limit of the maximum iteration number or the error precision.
Compared with the prior art, the scheme makes up the defect of asynchronous communication based on the topic of ROS, the event triggering condition is added into the synchronous algorithm, the algorithm is designed into the asynchronous algorithm, the final convergence result and the theoretical optimal solution error are reduced to 1%, so that multiple robots are separated from centralized control, each robot runs an independent program by adopting real distributed control, and the optimal solution is converged under the asynchronous communication environment. The scheme guarantees correct algorithm convergence on the premise of ros-based asynchronous high-efficiency communication, avoids the blocking behavior of a synchronous algorithm while maintaining the formation, and enables the algorithm convergence to be faster and the real-time performance of the robot system to be higher.
Drawings
FIG. 1 is a flow chart of the steps of the algorithm of the present invention;
FIG. 2 is a final effect diagram of a ros platform based ground enclosing trolley for an unmanned aerial vehicle in a test example of the invention;
fig. 3 is a graph of a robot position coordinate set (i.e., an equality constraint) obtained by different algorithms according to a test example of the present invention, along with the number of iterations.
Detailed Description
The present invention will be further illustrated with reference to the accompanying drawings and specific embodiments, which are to be understood as merely illustrative of the invention and not as limiting the scope of the invention.
Example 1
An ROS platform-oriented multi-robot formation algorithm based on distributed asynchronous optimization has the characteristics that each robot independently runs own programs and asynchronous communication, and for a system with n robots, the specific implementation steps of the n robots are shown in FIG. 1 for the i 1 st, 2 nd, 3 th.
S1, determining a formation and establishing a model according to the target to be enclosed, and establishing a local loss function, a constraint condition and a virtual initial position, wherein the method comprises the following specific steps:
the objective function and constraint conditions of the ith robot are shown
min f i (x i )=||x i -s i || 2 (1)
s.t x i ∈Ω i (2)
x i ∈S i (3)
Wherein x i ∈R 2 Is the position coordinate of the robot i, also the parameter to be optimized, s i ∈R 2 Is the best observation position of the robot i; objective function (1) to optimize robot position x i To the optimal observation position s i Is the smallest.
Constraint (2) indicates that robot i needs to be in its safe areax i ∈Ω i ={x i ∈R 2 ||x i -z i ∣≤r 1 In which z is i ∈R 2 Robot i safety Range center, r 2 Is an observation region S i The distance from the center to any boundary, i.e., half the side length of the square region. Constraint (3) indicates that robot i needs to be in its observation region S i ,x i ∈S i ={x i ∈R 2 ||x i -s i ∣<r 2 },r 2 For observation region S i The distance from the center to any border, i.e. half the side length of the square area. To satisfy the constraint conditions, the robot i needs to be in two square areas X i ,S i By solving the inequality, the intersection of Wherein, the constraints (2) and (3) can be converted into linear inequality constraints (6) and (7). The constraint condition (4) indicates that the position centers of all the robots are coincided with the target to be blocked, and d belongs to R 2 Indicating the location of the object to be occluded. The final optimization problem can be transferred to the following constrained optimization problem:
min f i (x i )=||x i -s i || 2 (5)
s2, judging whether the event trigger is satisfied by comparing whether the current update progress of the robot is slower than that of other nodes, wherein the event trigger conditions of the robot i are as follows:
k i for the local update times of the robot i, when the event trigger condition is satisfied, the robot i performs local update. If tau is 0, the event triggering of the robot i means that the robot i is the robot with the least number of updates in the current neighbor, and the idea of the event triggering is that if the number of updates of the robot i is more than that of other robots, it is described that the robot i has aggregated information of other robots for many times, and the behavior of aggregation for many times will cause the position average value of the robot system to shift towards the directions of other robots, and further cause the constraint condition to be unsatisfied, so that if the number of updates of the robot i is more than that of other agents, the robot i needs to stop to wait for other agents.
S3, local updating is carried out according to the updating condition of the robot i
Initializing the robot i parameter x i =z i ,k i =0,v i =(0,0,0,0),M i (1, -1), wherein the requirements are satisfiedWhen an event triggersWhen the condition is satisfied, updating according to the following formula:
k i =k i +1 (12) wherein I 2 Is an identity matrix of 2 x 2,is kronecker product, A ═ a ij ] m*m A is an adjacency matrix of the robot, when i ≠ j ij >0 if and only if the robot i sends j information, i j, (x) + =max{0,x}。v i is a lagrange multiplier. When the event trigger condition is triggered, the robot i calculates the gradient of the target function and a feasible gradient direction, moves towards the positive direction of the feasible gradient direction, then collects the information of the neighbor, moves towards the negative direction of the feasible gradient of the neighbor, and finally updates the Lagrange multiplier of the robot i. The neighbor's information is stored in a buffer (the old information is overwritten if there is new information).
S4, judging whether the termination condition is satisfied, if the termination condition has the maximum iteration number upper limit and the error precision, terminating the algorithm and outputting the final position x i If not, returning to the third step
Examples of tests:
taking a multi-robot ground enclosing trolley as an example, the ROS platform-oriented multi-robot formation algorithm based on distributed asynchronous optimization specifically comprises the following steps:
the first step is to initialize parameters and establish a robot model. In the experiment, the coordinates of the object to be blocked are d (15-15), three robots are adopted (n is 3), and finally the object to be blocked is blocked into a triangle.
Initializing a virtual initial position of the robot: z is a radical of 1 =(-65,-30),z 2 =(10,45),z 3 =(100,-60),Optimum observation position s of robot 1 =(-20,-30),s 2 =(10,5),s 3 =(55,-50),r 1 =30,r 2 =20。
Adjacency matrix a ═ a of robot system ij ] 3×3
Initializing other parameters k of the model 1 =k 2 =k 3 =,η=0.05,τ=1,v 1 =v 2 =v 3 =(0,0,0,0)。
For the ith robot, the objective function is f i (x i )=||x i -s i || 2 With the constraint condition of
And secondly, judging whether the event trigger is met by comparing whether the current updating progress of the robot is slower than that of other nodes or not, wherein the judgment is shown in the following formula:
wherein k is i And i is 1,2 and 3, the update times of the robot i, and the times of triggering the robot i event triggering condition. If the number of times of updating the robot τ is more than that of other robots, it is described that the robot i has aggregated information of other robots for many times, and the behavior of aggregation for many times may cause the position average value of the robot system to shift to the directions of other robots, thereby causing the constraint condition to be unsatisfied.
τ is an event triggering condition parameter, the larger τ represents that the event triggering condition has higher tolerance to delay, the larger τ is, the easier event triggering of the robot i is, the more frequent the robot is updated, and the convergence is faster, but the error of the convergence result is larger due to the higher tolerance to delay.
Therefore, τ can be increased if convergence speed is sought, and τ can be decreased if convergence accuracy is sought, τ ∈ N. The user can adjust τ to make the accuracy/speed tradeoff.
And thirdly, returning to the second step if the event trigger condition is not met for the ith robot, and updating the intelligent agent if the event trigger condition is met, wherein the specific updating steps are as follows:
When the event triggering condition is triggered, the robot i calculates the gradient of the target function and a feasible gradient direction, moves towards the negative direction of the feasible gradient direction, then collects the information of the neighbor and moves towards the positive direction of the feasible gradient of the neighbor, thereby keeping the center of the robot system and the target to be enclosed the same.
c. Finally updating the times k for the self i ,k i =k i +1。
After the local update is finished, updating the self number of times k i Local directionAnd sending to the neighbor.
Fourthly, judging whether a termination condition is met, wherein the termination condition has the maximum iteration number upper limit and the error precision, and if one is met, terminating the algorithm and outputting a final position x i And if not, returning to the third step.
As shown in fig. 2, fig. 2 is a final effect diagram of the ground blocking trolley for the unmanned aerial vehicle based on the ros platform, the star shape is the initial position of the unmanned aerial vehicle, the solid round shape is the final position of the unmanned aerial vehicle, i.e. the final output of the algorithm, the solid line hollow round area is the safety position of the unmanned aerial vehicle, the dotted line hollow circle center area is the observation area of the unmanned aerial vehicle, "×" is the target to be blocked, s is the target to be blocked, and 1 ,s 2 ,s 3 to observe the center, FIG. 2 graphically illustrates the model and the entire optimization process.
In the test example, a synchronous algorithm syn, an asynchronous algorithm asy and an algorithm asy-dc for adding an event trigger mechanism in the case are respectively used for testing, and a graph of a finally obtained robot position coordinate set (namely an equality constraint condition) along with the iteration times is shown in fig. 3, so that the synchronous algorithm can be well maintained at xy1+ xy2+ xy3 which is 3 × dy-45, and the ideal position coordinate set is the same as the synchronous algorithm which is-45; the arithmetic final position sum xy1+ xy2+ xy3 of the asynchronous communication based on ros without an event trigger mechanism is-71.2, and the error is up to 60%, while the arithmetic final position sum xy1+ xy2+ xy3 of the asynchronous arithmetic position sum with the event trigger compensation mechanism provided by the invention is-46.1, and the error is reduced to 2%. Therefore, on the premise of ros-based asynchronous high-efficiency communication, the algorithm ensures correct algorithm convergence, avoids the blocking behavior of a synchronous algorithm while maintaining the formation, and enables the algorithm convergence to be faster and the real-time performance of the robot system to be higher.
It should be noted that the above-mentioned contents only illustrate the technical idea of the present invention, and the protection scope of the present invention is not limited thereby, and it is obvious to those skilled in the art that several modifications and decorations can be made without departing from the principle of the present invention, and these modifications and decorations fall within the protection scope of the claims of the present invention.
Claims (5)
1. A ROS platform-oriented multi-robot formation algorithm based on distributed asynchronous optimization is applied to a system of n robots (1, 2, 3), and is characterized by comprising the following steps:
s1: determining a formation according to the position of a target to be enclosed and establishing a distributed optimization model, wherein the objective function of the distributed optimization model is the position x of the optimization robot i To the optimal observation position s i Is the smallest distance, i.e.:
minf i (x i )=||x i -s i || 2
wherein x is i ∈R 2 Is the position coordinates of robot i; s i ∈R 2 Is the best observation position of the robot i;
the constraints of the distributed optimization model at least comprise: robot i needs to be in its safe area Ω i And observe the region S i The centers of the positions of the inner robot and all the robots are coincided with the target to be blocked, namely:
s.tx i ∈Ω i
x i ∈S i
wherein d ∈ R 2 Representing the position of the target to be enclosed;
s2: judging whether event triggering is met by comparing whether the current updating progress of the robot is slower than other nodes or not, if the updating times of the robot i are more than that of other nodes, stopping the robot i to wait for other agents; when the event trigger condition is satisfied, the robot i performs local update, and continues to step S3; the event triggering conditions of the robot i are as follows:
wherein k is i The number of local updates for robot i; k is a radical of j Local update times for robot j; tau is an event triggering condition parameter; n is a radical of i A neighbor set of a robot i;
s3: and local updating according to the updating condition of the robot i, wherein the specific formula is as follows:
k i =k i +1
where eta is the step length, a ij As a weight of the neighbor(s),is a gradient, v i Is the Lagrangian constant; i is 2 Is an identity matrix of 2 x 2,is kronecker product, A ═ a ij ] m*m A is an adjacency matrix of the robot, when i ≠ j ij >0 if and only if the robot i sends j information, i j,
s4: judging whether a termination condition is met, if so, terminating the algorithm and outputting a final position x i (ii) a If not, the step is returned to the step S3 to repeat the steps, and the termination condition is set as the upper limit of the maximum iteration number or the error precision.
2. The ROS-oriented platform multi-robot formation algorithm based on distributed asynchronous optimization of claim 1, wherein: in the step S1, the robot i needs to be in the safety area Ω i And observe the region S i In the interior of said container body,
x i ∈Ω i ={x i ∈R 2 ||x i -z i ∣≤r 1 },
x i ∈S i ={x i ∈R 2 ||x i -s i ∣<r 2 }
wherein z is i ∈R 2 Is the robot i safety range center; r is 1 For a safe region omega i The distance from the center to any boundary, namely half of the side length of the square area; r is 2 Is an observation region S i The distance from the center to any boundary, i.e., half the side length of the square region.
4. the ROS-oriented platform multi-robot formation algorithm based on distributed asynchronous optimization of claim 2 or 3, characterized in that: in step S2, the convergence speed or accuracy of the model can be adjusted by changing the value of τ, and τ can be increased if the convergence speed is pursued; if convergence accuracy is sought, τ ∈ n can be reduced.
5. The ROS-oriented platform multi-robot formation algorithm based on distributed asynchronous optimization of claim 4, wherein: in step S3, when the event trigger condition is satisfied, the robot i calculates the gradient of the target function and a feasible gradient direction, and moves in the negative direction of the feasible gradient direction, and then collects the information of the neighbor and moves in the positive direction of the feasible gradient of the neighbor, so as to keep the center of the robot system and the target to be blocked the same.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210520624.0A CN114995397A (en) | 2022-05-13 | 2022-05-13 | ROS (reactive oxygen species) platform-oriented multi-robot formation algorithm based on distributed asynchronous optimization |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210520624.0A CN114995397A (en) | 2022-05-13 | 2022-05-13 | ROS (reactive oxygen species) platform-oriented multi-robot formation algorithm based on distributed asynchronous optimization |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114995397A true CN114995397A (en) | 2022-09-02 |
Family
ID=83027628
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210520624.0A Pending CN114995397A (en) | 2022-05-13 | 2022-05-13 | ROS (reactive oxygen species) platform-oriented multi-robot formation algorithm based on distributed asynchronous optimization |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114995397A (en) |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102331711A (en) * | 2011-08-12 | 2012-01-25 | 江苏合成物联网科技有限公司 | Formation control method for mobile autonomous robots |
US20200167437A1 (en) * | 2018-11-27 | 2020-05-28 | Amazon Technologies, Inc. | Simulation orchestration for training reinforcement learning models |
CN112348157A (en) * | 2020-11-18 | 2021-02-09 | 中国空间技术研究院 | Collaborative optimization method and device for formation configuration of multi-agent system |
CN113642243A (en) * | 2021-08-23 | 2021-11-12 | 香港中文大学(深圳) | Multi-robot deep reinforcement learning system, training method, device and medium |
CN114063614A (en) * | 2021-10-27 | 2022-02-18 | 山东科技大学 | Rapid convergence method for mixed formation of multi-small robot system based on airborne monocular visual perception |
WO2022088471A1 (en) * | 2020-10-28 | 2022-05-05 | 江南大学 | Variable-batch-length iterative learning optimization control method for mobile robot |
-
2022
- 2022-05-13 CN CN202210520624.0A patent/CN114995397A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102331711A (en) * | 2011-08-12 | 2012-01-25 | 江苏合成物联网科技有限公司 | Formation control method for mobile autonomous robots |
US20200167437A1 (en) * | 2018-11-27 | 2020-05-28 | Amazon Technologies, Inc. | Simulation orchestration for training reinforcement learning models |
WO2022088471A1 (en) * | 2020-10-28 | 2022-05-05 | 江南大学 | Variable-batch-length iterative learning optimization control method for mobile robot |
CN112348157A (en) * | 2020-11-18 | 2021-02-09 | 中国空间技术研究院 | Collaborative optimization method and device for formation configuration of multi-agent system |
CN113642243A (en) * | 2021-08-23 | 2021-11-12 | 香港中文大学(深圳) | Multi-robot deep reinforcement learning system, training method, device and medium |
CN114063614A (en) * | 2021-10-27 | 2022-02-18 | 山东科技大学 | Rapid convergence method for mixed formation of multi-small robot system based on airborne monocular visual perception |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108932216B (en) | Robot inverse kinematics solving method based on particle swarm optimization algorithm | |
CN109765893A (en) | Method for planning path for mobile robot based on whale optimization algorithm | |
CN105159096B (en) | A kind of redundancy space manipulator joint moment optimization method based on particle cluster algorithm | |
CN110076783B (en) | Planar under-actuated mechanical arm position control method based on fuzzy control and Lyapunov function | |
JP5774361B2 (en) | Trajectory planning method, trajectory planning system, and trajectory planning / control system | |
CN112578804B (en) | Four-rotor aircraft formation sliding mode control method based on event trigger mechanism | |
CN104407619A (en) | Method enabling multiple unmanned aerial vehicles to reach multiple targets simultaneously under uncertain environments | |
CN113342037B (en) | Multi-rotor unmanned aerial vehicle time-varying formation control method and system with input saturation | |
CN112631335B (en) | Event triggering-based multi-quad-rotor unmanned aerial vehicle fixed time formation method | |
WO2021082709A1 (en) | Method and device for planning path of underwater dredging robot, robot, and storage medium | |
CN109262612B (en) | Under-actuated mechanical arm joint angle optimization method based on improved particle swarm optimization | |
CN112161627A (en) | Intelligent path planning method for fire-fighting robot | |
CN114169234A (en) | Scheduling optimization method and system for unmanned aerial vehicle-assisted mobile edge calculation | |
CN114265318A (en) | Cooperative robot trajectory tracking method based on sliding mode control and fuzzy algorithm | |
CN108279704B (en) | Cooperative control strategy for multiple unmanned aerial vehicles with increased communication links | |
CN111482969A (en) | Six-degree-of-freedom offset robot inverse solution method based on BAS algorithm | |
CN114995397A (en) | ROS (reactive oxygen species) platform-oriented multi-robot formation algorithm based on distributed asynchronous optimization | |
CN112016162A (en) | Four-rotor unmanned aerial vehicle PID controller parameter optimization method | |
CN116880197B (en) | Underwater robot operation track planning optimization method and optimization system based on multi-target multi-population backbone particle swarm optimization algorithm | |
Chemori et al. | A prediction‐based nonlinear controller for stabilization of a non‐minimum phase PVTOL aircraft | |
CN115981375B (en) | Design method of multi-unmanned aerial vehicle time-varying formation controller based on event triggering mechanism | |
CN117666616A (en) | High-speed aircraft cluster formation control method and system based on SAC reinforcement learning algorithm | |
CN115933415A (en) | Event trigger pulse-based multi-agent consistency cooperative control method | |
CN114564044B (en) | Unmanned aerial vehicle limited time formation control method triggered by input amplitude limiting event | |
CN116300881A (en) | Power inspection robot path planning method based on improved BIT and artificial potential field method |
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 |