CN115373266A - Rope-constrained multi-agent tension prediction and cooperative control method - Google Patents
Rope-constrained multi-agent tension prediction and cooperative control method Download PDFInfo
- Publication number
- CN115373266A CN115373266A CN202210962852.3A CN202210962852A CN115373266A CN 115373266 A CN115373266 A CN 115373266A CN 202210962852 A CN202210962852 A CN 202210962852A CN 115373266 A CN115373266 A CN 115373266A
- Authority
- CN
- China
- Prior art keywords
- agent
- neural network
- formation
- tension
- rope
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 33
- 230000015572 biosynthetic process Effects 0.000 claims abstract description 51
- 238000013528 artificial neural network Methods 0.000 claims abstract description 43
- 238000002474 experimental method Methods 0.000 claims abstract description 24
- 238000012549 training Methods 0.000 claims abstract description 17
- 238000005457 optimization Methods 0.000 claims abstract description 8
- 239000003795 chemical substances by application Substances 0.000 claims description 103
- 230000006870 function Effects 0.000 claims description 15
- 210000002569 neuron Anatomy 0.000 claims description 12
- 230000004913 activation Effects 0.000 claims description 3
- 238000011478 gradient descent method Methods 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 3
- 238000013507 mapping Methods 0.000 description 5
- 230000001133 acceleration Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 230000008520 organization Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- AMGNHZVUZWILSB-UHFFFAOYSA-N 1,2-bis(2-chloroethylsulfanyl)ethane Chemical compound ClCCSCCSCCCl AMGNHZVUZWILSB-UHFFFAOYSA-N 0.000 description 1
- 241000283690 Bos taurus Species 0.000 description 1
- 241000257303 Hymenoptera Species 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000001568 sexual effect Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 239000000725 suspension Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Feedback Control In General (AREA)
Abstract
The invention relates to a method for acquiring the whole-process state data of a rope tying restraint body in a multi-agent system under a complete task in a demonstration experiment environment, utilizing state data set training to obtain the prediction of a model tension uncertainty item according to the approaching characteristic of a RBF neural network, deploying the prediction into the multi-agent system, perfecting a system model, and completing a formation task through a control algorithm, thereby solving the technical problem that the existing multi-agent system cannot carry out accurate formation. The invention establishes a closed loop flow of demonstrative experiment of a multi-agent formation system, RBF neural network training, tension uncertainty predicted value deployment and repeated experiment of the multi-agent formation system, utilizes the training result of experimental data in the RBF neural network, continuously designs an optimization controller, and can achieve the optimal control effect as far as possible compared with the prior formation control technology.
Description
Technical Field
The invention belongs to the field of multi-agent formation control, and relates to a rope-constrained multi-agent tension prediction and cooperative control method, in particular to a formation control method of multi-agent under rope constraint.
Background
The multi-agent formation belongs to a multi-agent formation system, and an agent generally refers to a physical or abstract entity which can sense the environment of the agent and correctly call the knowledge of the agent to make a proper response to the environment. A multi-agent system generally refers to a complex system composed of multiple agents and their corresponding organization rules and information interaction protocols, and capable of accomplishing specific tasks. The organization rules determine the connection relation between the agents, and the information interaction protocol is used for determining and updating the states of the agents. Compared with a large number of multi-intelligent-agent system examples in the real world, such as the cooperation of multiple ants for carrying food, the organized migration of cattle, the formation of clusters of birds for flying and the like, the cooperation of the multi-intelligent-agent system can complete more complex tasks with lower cost, the traditional control theory facing a single object can hardly meet the actual control requirement, and the multi-intelligent-agent system has obvious advantages due to the characteristics of strong function, flexible structure, strong expandability and the like.
Formation control means that a team consisting of a plurality of agents keeps a predetermined collective formation shape and avoids obstacles in the process of moving to a specific target. Generally speaking, the formation control implements the clustering behavior of a multi-agent system with the help of local neighbor agent information of agents, thereby solving a global task. The multi-agent formation control has wide application prospect in various fields of military, aerospace, industry and the like, such as satellite navigation, robot control, search and rescue and the like.
The tether restraint is a restraint of an object by a tether such as a tether rope, a chain, or a belt. In a multi-agent system, compared to rigid constraints, tether constraints can keep the constrained multi-agent within a certain range of motion space, allowing the constrained multi-agent to have a certain degree of displacement. Even under the condition of ensuring the tensioning of a rope system, the traditional multi-intelligent-agent rope system formation control method still needs an accurate system dynamics and environment model, but the problems of uncertainty, inaccuracy, nonlinearity, complexity, time-varying property and the like of the system model are further complicated due to the characteristics of rope system constraint, and the uncertainty of the tension of the rope system causes that the existing multi-intelligent-agent formation control method cannot accurately control the multi-intelligent agents, so that the state dimension of the multi-intelligent agents is changed, and the formation configuration is disordered.
Therefore, a new multi-agent formation control method under the constraint of a rope is needed to solve the technical problem that the existing multi-agent cannot be accurately formed.
Disclosure of Invention
Technical problem to be solved
In order to avoid the defects of the prior art, the invention provides a tension prediction and cooperative control method of rope restraint multi-agent, which comprises the steps of obtaining whole-course state data of a rope restraint in a multi-agent system under a complete task in a demonstration experiment environment, training by using state data sets according to the approaching characteristic of an RBF (radial basis function) neural network to obtain prediction of model tension uncertainty, deploying the prediction into the multi-agent system, perfecting a system model, and completing a formation task through a control algorithm, thereby solving the technical problem that the existing multi-agent cannot perform accurate formation.
Technical scheme
A tension prediction and cooperative control method for rope-series constraint multi-agent is characterized by comprising the following steps:
step 1: aiming at multi-agent cooperative formation, establishing a multi-agent formation dynamics model containing flexible constraint uncertain nonlinear items:
wherein X represents the state quantity of the agent, u represents the control force, F cable The method is characterized in that the method is a rope system restraint uncertain item, namely rope system restraint tension, and d represents unknown external interference of the system;
and 2, step: for multi-agent cooperative formation control, a sliding mode controller with robust performance is preliminarily designed:
wherein x represents state quantity, g (x) is coefficient term of control force in the model, D is external disturbance upper bound, beta, p, q and eta are control parameters, beta is more than 0, p and q are positive odd numbers,
η＞0，
is a slip form surface;
and step 3: according to a controller expression and experimental conditions, after a marginalized external interference upper bound D and other control parameters are set, performing a demonstrative experiment on the multi-agent formation system to obtain demonstrative experiment state data of each agent in the multi-agent formation system;
and 4, step 4: using the state data of each intelligent agent in the demonstration experiment as the input vector of the neural network
E represents a state quantity error;
and 5: designing a system model rope system constraint tension uncertainty expected expression:
wherein, the rope tension T = [ T ] 1 T 2 T 3 ] T Swing angle alpha of tether line of system i 、β i . Calculated expected value of tension uncertainty
Outputting tag value f as neural network * ；
Step 6: inputting the gamma of the step 4 into an RBF neural network, and taking the f as the input * Training the RBF neural network for a label value to obtain a trained Gaussian radial basis function central point C, a trained variance sigma and a network parameter W from a network hidden layer to an output layer * 、b * ；
And 7: the center point C, the variance sigma and the network parameter W of the trained Gaussian radial basis function * 、b * Deploying the active compensation term into a new multi-unmanned aerial vehicle rope system load dynamic model to obtain an active compensation term predicted value
Wherein the content of the first and second substances,
real-time RBF neural network input derived for multi-agent real-time status data, E realtime Representing the error of the real-time state quantity of the intelligent agent;
and 8: predicting active compensation term
Added to the original controller u 0 And updating the optimization controller:
the optimized controller u 1 Deployment to an experimental model
And (3) carrying out a new demonstration experiment, and iterating the steps 4-8 until the experimental state data of the multiple agents meet the formation configuration optimization target:
||X i -X j ||＞d safety
X i →X d
wherein, X i ,X j Representing different state quantities of the agent, d safety Is a safe distance, X d Is the desired state quantity.
The RBF neural network is a three-layer neural network, wherein on a hidden layer, the activation function of each neuron is a Gaussian radial basis function
And the central point C, the weight value matrix from the hidden layer to the output layer is W, and the bias item b is used.
The RBF neural network and
as a loss function, m is the number of samples of the experimentally collected data set, and the network parameter update algorithm is
Gradient descent method, where θ is the neural network parameter and learning _ rate is the learning rate.
The number of the neurons in the hidden layer is according to an empirical formula:
wherein N is s Is the number of data set samples, N i Is the number of neurons in the input layer, N o Is the number of neurons in the output layer, and a is the regulation coefficient.
The adjusting coefficient a is 2-10.
Advantageous effects
The invention provides a tension prediction and cooperative control method of rope constraint multi-agent, which is characterized in that the whole-course state data of a rope constraint body in a multi-agent system under a complete task in a demonstration experiment environment is obtained, the prediction of model tension uncertainty is obtained by utilizing state data set training according to the approaching characteristic of a Radial Basis Function (RBF) neural network, the prediction is deployed in the multi-agent system, a system model is perfected, and a formation task is completed through a control algorithm, so that the technical problem that the existing multi-agent system cannot perform accurate formation is solved.
Compared with the prior art, the invention has the beneficial effects that:
(1) The invention considers the concrete influence of the uncertain rope restraint tension on the formation model in the task executed by the multi-agent, and provides a multi-agent dynamic model containing rope restraint force;
(2) In the invention, the expected mapping relation between the state quantity of each intelligent agent and the tension uncertainty item is solved by utilizing the geometrical constraint and the dynamic characteristic of the rope system in the multi-intelligent-agent formation system, and compared with a method for performing anti-disturbance control by taking the tension uncertainty item brought by the rope system constraint as interference in the traditional control task, the method is based on an expected expression of the tension uncertainty item;
(3) The RBF neural network with approximation capability is introduced to predict the rope constraint tension uncertainty in the multi-agent formation system, and compared with the traditional passive compensation type control method, the method has the advantages of higher timeliness and better control effect;
(4) The invention establishes a closed loop process of demonstration experiment of a multi-agent formation system, RBF neural network training, tension uncertainty predicted value deployment and multi-agent formation system repeated experiment, utilizes the training result of experimental data in the RBF neural network, and continuously designs an optimization controller.
Drawings
Fig. 1 is a schematic diagram of a system for connecting three drones to a load formation system through tethers according to an embodiment of the invention;
FIG. 2 is a schematic diagram of the RBF neural network structure of the present invention;
fig. 3 is a schematic view of an RBF neural network training and control structure of a load formation system connected by three drones through tethers in an embodiment of the present invention;
FIG. 4 is a schematic flow chart of a tethered multi-agent formation control method and system of the present invention.
Detailed Description
The invention will now be further described with reference to the following examples and drawings:
a method of controlling multi-agent formation, comprising the steps of:
according to the planning target of the multi-agent formation optimization configuration, a controller is preliminarily designed to perform a demonstration experiment on the multi-agent system containing the rope constraint;
acquiring demonstrative experiment state data of each agent in the multiple agents;
designing an expected expression of a system model rope system constraint tension uncertainty item, and establishing a mapping relation from a state quantity to the rope system constraint tension uncertainty item;
the multi-agent formation optimization configuration goal should meet:
||X i -X j ||＞d safety
X i →X d
wherein, X i ,X j Representing different state quantities of the agent, d safety Is a safe distance, X d Is a desired state quantity;
the state data is used as input, and the rope system constraint tension uncertainty expected value is obtained through solving through the mapping relation between the state data and the rope system constraint tension uncertainty;
constructing an RBF neural network which takes the state data of each intelligent agent as input quantity and takes the tension uncertainty expected value obtained by calculation as an output label according to the state data;
training an RBF neural network to obtain a mapping relation between the input state quantity and an uncertain item of rope system restraint tension;
acquiring real-time state data of a plurality of intelligent agents to be controlled in formation, deploying RBF neural network training results into a system model, inputting the real-time state data serving as a network, and solving through the RBF neural network training results to obtain real-time rope system constraint tension uncertainty in the multi-intelligent agent system;
designing a multi-intelligent controller with active compensation capability by combining the obtained uncertain rope restraint tension items;
and performing a demonstration experiment on the formation of the multi-agent containing rope constraint, and repeating the steps until the state data of the multi-agent reaches an ideal value.
The research object is a rope-bound multi-agent formation system, and the dynamic model can be summarized as follows:
wherein X represents the state quantity of the agent, u represents the control force, F cable Is the tether restraint term (i.e., tether restraint tension), and d represents the unknown external disturbance of the system.
The preliminary controller in the multi-agent system including the binding force of the rope system takes a sliding mode controller with better robust performance as an example, and a designed control force calculation formula is as follows:
wherein x represents state quantity, g (x) is coefficient term of control force in the model, D is external disturbance upper bound, beta, p, q and eta are control parameters, beta is more than 0, p and q are positive odd numbers,
η＞0，
is a slip form surface.
Desired expression f for the rope system constraint tension uncertainty term * (Γ) determined by the geometrical constraints and the dynamics of the specific multi-agent formation configuration;
the state data comprises control force data and position, speed and acceleration state information of each agent;
the training result of the RBF neural network is the mapping relation between the state input and the label, and the formula is expressed as follows:
f＝W T ·h(Γ)+b
h (Γ) is a gaussian radial basis function,
is the state input vector, C is the center point, σ is the variance, W, b are the weight values and bias terms of the neural network output layer, and f is the neural network output value.
Obtaining ideal network parameters W after training RBF neural network * 、b * The method is characterized in that the method is deployed into a multi-agent formation dynamics model, a new controller is designed by combining system real-time state data and a control theory, and the formula is expressed as follows:
will u 1 And performing a new turn of demonstrative experiment on the system as a new multi-agent controller, obtaining the experimental state data again, and continuously iterating until an ideal state is achieved.
The embodiments of the invention will be described in detail below with reference to the drawings, but the invention can be implemented in many different ways as defined and covered by the claims.
Taking a system for cooperatively carrying a tether load by multiple unmanned aerial vehicles as an example, the method comprises the following specific implementation steps:
as shown in fig. 1, for a multi-drone tether suspension load system including tether constraints, firstly, task objectives are set as follows:
min(||X-X d || 2 )
wherein X is the state vector of the agent and the load, X d The task target can be set for the expected value according to different task requirements;
the rope system tether tension is taken as an external force to be brought into the dynamics analysis, and the multi-unmanned aerial vehicle formation dynamics model is obtained through analysis as follows:
wherein u is 1i 、u 2i 、u 2i 、u 4i Respectively representing the virtual control input of the ith agent, F cable Representing the tether tie-line tension received by the entire multi-agent system, W representing gust interference, and d representing unknown external interference of the system. By making a pair F cable In response to the control input u, the multi-agent is designed to operate in a different manner than the maneuver 1i 、u 2i 、u 2i 、u 4i Relationship to desired formation configuration.
The initial controller expression is designed as follows:
acquiring data such as flight state data of multiple unmanned aerial vehicles and tether hanging load positions in a demonstration experiment scene;
the state data comprises position deviation data and speed deviation data of target points and measuring points of all intelligent agents and loads;
the included angle data between the rope tied rope and each intelligent body is a swing angle alpha capable of determining the relative position between the intelligent body and the load i 、β i Setting the length of the tether to L r The relative position relationship between the load P and each agent is as follows:
wherein the definition is from P to a single agent Q i The unit direction vector of (a) is: rho i ＝[cos(β i )cos(α i ),cos(β i )sin(α i ),sin(β i )] T ，
ξ p The position of the agent and the load in the ground coordinate system. Obtaining the swing angle alpha of the rope tying rope of the system by inverse solution by utilizing the relative relation between the multi-agent and the load space i 、β i ；
Analyzing the tether hanging load P, calculating and solving the tether tension T = [ T ] according to the following formula by using the state data of each intelligent object acquired by the demonstrative experiment 1 T 2 T 3 ] T ：
Wherein, a x 、a y 、a z Is tether load acceleration, m p Is the load mass.
Obtaining the uncertain item expected value of the rope restraint tension through the steps, wherein the formula is expressed as follows:
writing each intelligent state data in the demonstration experiment into a neural network input vector
E represents the error of the state quantity, and the calculated tension uncertainty expected value is used as the output label value f of the neural network * ；
RBF neural network structure As shown in FIG. 2, the activation function of each neuron on the hidden layer of the three-layer neural network is Gaussian radial basis function
And W is a weight value matrix from the hidden layer to the output layer, and b is a bias item. Hidden layer neuron number basis demonstrationThe size of the number of samples of the data set obtained by the sexual experiment can be determined according to an empirical formula:
wherein N is s Is the number of data set samples, N i Is the number of neurons in the input layer, N o The number of neurons in an output layer is, a is a regulation coefficient and is generally 2-10;
training the RBF neural network to
As a loss function, m is the number of samples of the experimentally collected data set, and the network parameter update algorithm is
Gradient descent method, where θ is neural network parameter and learning _ rate is learning rate;
the central point C and the variance sigma of the Gaussian radial basis function obtained from the RBF neural network training result, the weight value W from the network hidden layer to the output layer and the bias item b are deployed in a new multi-unmanned aerial vehicle tether load dynamic model to obtain the predicted value of the active compensation item
Updating the optimal controller expression:
and deploying the optimized controller into an experimental model, carrying out a new demonstration experiment, and iterating the steps until the load state quantities of the multiple unmanned aerial vehicles and the tether reach an expected target.
Although the present invention has been described with reference to the preferred embodiments, it is not intended to limit the present invention, and those skilled in the art can make variations and modifications of the present invention without departing from the spirit and scope of the present invention by using the methods and technical contents disclosed above.
Claims (5)
1. A tension prediction and cooperative control method for rope-constrained multi-agent is characterized by comprising the following steps:
step 1: aiming at multi-agent cooperative formation, a multi-agent formation dynamics model containing flexible constraint uncertain nonlinear items is established:
wherein X represents the state quantity of the agent, u represents the control force, F cable The method is characterized in that the method is a rope system restraint uncertain item, namely rope system restraint tension, and d represents unknown external interference of the system;
step 2: for multi-agent cooperative formation control, a sliding mode controller with robust performance is preliminarily designed:
wherein x represents state quantity, g (x) is coefficient term of control force in the model, D is external disturbance upper bound, beta, p, q and eta are control parameters, beta is more than 0, p and q are positive odd numbers,
is a slip form surface;
and step 3: setting an upper bound D of external interference with margin and other control parameters according to a controller expression and experimental conditions, and then performing a demonstrative experiment on the multi-agent formation system to obtain demonstrative experiment state data of each agent in the multi-agent;
and 4, step 4: using the state data of each intelligent agent in the demonstration experiment as the input vector of the neural network
E represents a state quantity error;
and 5: designing an expected expression of a system model tether restraint tension uncertainty term:
wherein, the rope tension T = [ T ] 1 T 2 T 3 ] T Swing angle alpha of tethered tether of system i 、β i . Calculated expected value of tension uncertainty
Outputting tag value f as neural network * ；
Step 6: inputting the Gamma of the step 4 to an RBF neural network, and using the F * Training the RBF neural network for a label value to obtain a trained Gaussian radial basis function central point C, a trained variance sigma and a network parameter W from a network hidden layer to an output layer * 、b * ；
And 7: the center point C, the variance sigma and the network parameter W of the trained Gaussian radial basis function * 、b * Deploying the active compensation term in a new multi-unmanned aerial vehicle tether load dynamic model to obtain an active compensation term predicted value
Wherein, the first and the second end of the pipe are connected with each other,
real-time RBF neural network input derived for multi-agent real-time status data, E realtime Representing the error of the real-time state quantity of the intelligent agent;
and step 8: predicting active compensation term
Added to the original controller u 0 And updating the optimization controller:
the optimized controller u 1 Deployment to an experimental model
And (3) carrying out a new demonstration experiment, and iterating the steps 4-8 until the experimental state data of the multiple agents meet the formation configuration optimization target:
||X i -X j ||＞d safety
X i →X d
wherein, X i ,X j Representing different state quantities of the agent, d safety Is a safe distance, X d Is the desired state quantity.
2. The tethered constrained multi-agent tension prediction and cooperative control method of claim 1, wherein: the RBF neural network is a three-layer neural network, wherein on a hidden layer, the activation function of each neuron is a Gaussian radial basis function
And the central point C, the weight value matrix from the hidden layer to the output layer is W, and the bias item b is used.
3. The tethered multi-agent tension prediction and cooperative control method of claim 1, wherein: the RBF neural network and
as a loss function, m is the number of samples of the experimental collection data set, and the network parameter updating algorithm is
Gradient descent method, where θ is the neural network parameter and learning _ rate is the learning rate.
4. The tethered multi-agent tension prediction and cooperative control method of claim 1, wherein: the number of the neurons in the hidden layer is according to an empirical formula:
wherein N is s Is the number of data set samples, N i Is the number of neurons in the input layer, N o Is the number of neurons in the output layer, and a is the regulation coefficient.
5. The tethered constrained multi-agent tension prediction and cooperative control method of claim 4, wherein: the adjusting coefficient a is 2-10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210962852.3A CN115373266B (en) | 2022-08-11 | 2022-08-11 | Tension prediction and cooperative control method for multiple agents in tethered constraint |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210962852.3A CN115373266B (en) | 2022-08-11 | 2022-08-11 | Tension prediction and cooperative control method for multiple agents in tethered constraint |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115373266A true CN115373266A (en) | 2022-11-22 |
| CN115373266B CN115373266B (en) | 2024-04-30 |
Family
ID=84065491
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210962852.3A Active CN115373266B (en) | 2022-08-11 | 2022-08-11 | Tension prediction and cooperative control method for multiple agents in tethered constraint |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115373266B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110442129A (en) * | 2019-07-26 | 2019-11-12 | 中南大学 | A kind of control method and system that multiple agent is formed into columns |
| CN114167732A (en) * | 2021-12-13 | 2022-03-11 | 西北工业大学 | Coupling constraint multi-agent distributed robust nonlinear model prediction control method |
| CN114237291A (en) * | 2021-12-15 | 2022-03-25 | 西北工业大学 | Flexible constraint multi-unmanned aerial vehicle formation control method |
| JP2022106261A (en) * | 2021-01-06 | 2022-07-19 | 北京科技大学 | Control method for intelligent vehicle rank formation on two-dimensional plane |
-
2022
- 2022-08-11 CN CN202210962852.3A patent/CN115373266B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110442129A (en) * | 2019-07-26 | 2019-11-12 | 中南大学 | A kind of control method and system that multiple agent is formed into columns |
| JP2022106261A (en) * | 2021-01-06 | 2022-07-19 | 北京科技大学 | Control method for intelligent vehicle rank formation on two-dimensional plane |
| CN114167732A (en) * | 2021-12-13 | 2022-03-11 | 西北工业大学 | Coupling constraint multi-agent distributed robust nonlinear model prediction control method |
| CN114237291A (en) * | 2021-12-15 | 2022-03-25 | 西北工业大学 | Flexible constraint multi-unmanned aerial vehicle formation control method |
Non-Patent Citations (1)
| Title |
|---|
| 周合;张帆;黄攀峰;: "空间绳系编队的动力学及稳定展开控制研究", 宇航学报, no. 04, 30 April 2020 (2020-04-30) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115373266B (en) | 2024-04-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Kang et al. | Generalization through simulation: Integrating simulated and real data into deep reinforcement learning for vision-based autonomous flight | |
| Shi et al. | Multi pseudo Q-learning-based deterministic policy gradient for tracking control of autonomous underwater vehicles | |
| Cheng et al. | Concise deep reinforcement learning obstacle avoidance for underactuated unmanned marine vessels | |
| CN110806759B (en) | Aircraft route tracking method based on deep reinforcement learning | |
| Rueckert et al. | Learning inverse dynamics models in o (n) time with lstm networks | |
| CN109696830B (en) | Reinforced learning self-adaptive control method of small unmanned helicopter | |
| Ferdaus et al. | Towards the use of fuzzy logic systems in rotary wing unmanned aerial vehicle: a review | |
| Li et al. | Compensating delays and noises in motion control of autonomous electric vehicles by using deep learning and unscented Kalman predictor | |
| Anderson et al. | Challenging control problems | |
| Ishii et al. | An adaptive neural-net controller system for an underwater vehicle | |
| Nie et al. | Three-dimensional path-following control of a robotic airship with reinforcement learning | |
| Yang et al. | Horizontal trajectory control of stratospheric airships in wind field using Q-learning algorithm | |
| CN111880567A (en) | Fixed-wing unmanned aerial vehicle formation coordination control method and device based on deep reinforcement learning | |
| Kayacan et al. | Learning control of fixed-wing unmanned aerial vehicles using fuzzy neural networks | |
| CN112925347B (en) | Multi-unmanned aerial vehicle sliding mode formation control method based on data driving | |
| CN113759979B (en) | Event-driven-based online track planning method for unmanned aerial vehicle hanging system | |
| CN111753464A (en) | Unmanned helicopter model online learning method and system | |
| Goecks et al. | Control of morphing wing shapes with deep reinforcement learning | |
| Zhang et al. | A novel event-triggered robust neural formation control for USVs with the optimized leader–follower structure | |
| Soylu et al. | Dynamics and control of tethered underwater-manipulator systems | |
| CN107315348B (en) | Position and posture control method of unmanned underwater vehicle based on punishment type wavelet network | |
| Zhao et al. | Data-driven formation control for multiple heterogeneous vehicles in air–ground coordination | |
| CN114237268A (en) | Unmanned aerial vehicle strong robust attitude control method based on deep reinforcement learning | |
| Song et al. | Surface path tracking method of autonomous surface underwater vehicle based on deep reinforcement learning | |
| Li et al. | Spacecraft Relative Trajectory Planning Based on Meta-Learning |
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 |