CN108646550B - Multi-agent formation method based on behavior selection - Google Patents

Multi-agent formation method based on behavior selection Download PDF

Info

Publication number
CN108646550B
CN108646550B CN201810286164.3A CN201810286164A CN108646550B CN 108646550 B CN108646550 B CN 108646550B CN 201810286164 A CN201810286164 A CN 201810286164A CN 108646550 B CN108646550 B CN 108646550B
Authority
CN
China
Prior art keywords
agent
behavior
detection
basal ganglia
channel
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
Application number
CN201810286164.3A
Other languages
Chinese (zh)
Other versions
CN108646550A (en
Inventor
金贝
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jiangsu Jiangrong Intelligent Technology Co ltd
Original Assignee
Jiangsu Jiangrong Intelligent Technology Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Jiangsu Jiangrong Intelligent Technology Co ltd filed Critical Jiangsu Jiangrong Intelligent Technology Co ltd
Priority to CN201810286164.3A priority Critical patent/CN108646550B/en
Publication of CN108646550A publication Critical patent/CN108646550A/en
Application granted granted Critical
Publication of CN108646550B publication Critical patent/CN108646550B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a multi-agent formation method based on behavior selection, which comprises the following steps: step one, setting a detection profit calculation method related to the position; secondly, defining specific behaviors of a plurality of agents; determining the channel number of the basal ganglia, establishing a basal ganglia channel model and initializing relevant parameters; and step four, correcting the channel model parameters of the basal ganglia. Through the mode, the multi-agent formation method based on behavior selection introduces a dynamic agent formation method based on behavior selection, effectively overcomes the defects of low robustness, low reliability and the like in the existing formation detection method, enables the technical scheme of the application to have wider universality by opening parameters and various weight modification channels, introduces a target detection yield variable, and can integrally improve the benefit of the multi-agent for detection tasks.

Description

Multi-agent formation method based on behavior selection
Technical Field
The invention relates to the field of formation of intelligent agents, in particular to a multi-intelligent-agent formation method based on behavior selection.
Background
The intelligent agent can replace people to finish repetitive work in complex and dangerous environments, and social development is accelerated. In recent years, market environments put new demands on the functions of agents, agents need to complete more complex detection tasks and even operation tasks, but in the face of complex tasks which need to be completed in parallel with high efficiency, a single agent cannot be competent, and multi-agent cooperation is needed.
The intelligent agent formation control is a control method which can not only form a target formation but also adapt to specific environmental constraints when a plurality of intelligent agents reach the target formation. The existing common methods mainly comprise a method based on a pilot-follower method, a virtual structure method, a method based on behaviors, a method based on graph theory and a potential energy method.
The basic idea of the navigator-follower method is that one or more navigation intelligent agents can exist in a formation system formed by multiple intelligent agents, other non-navigation intelligent agents are following intelligent agents, and the following intelligent agents take the positions, relative distances and relative angles relative to the navigation intelligent agents as input control quantities, so that the relative positions of the following intelligent agents and the navigation intelligent agents approach target values infinitely. The control structure of the piloting following method is simple, but because no position feedback exists between the piloting intelligent agent and the following intelligent agent and the piloter intelligent agent is in single-point control, the situations that the intelligent agents fall behind and the like easily occur, and the robustness of the system is poor.
The basic idea of the virtual structure method is to consider a system composed of multiple agents as an imaginary rigid structure, and the coordinates of each agent in a reference coordinate system are unchanged, namely the relative positions of the agents are unchanged.
The basic idea of the behavior-based method is that each link of multi-agent formation is regarded as being formed by a plurality of basic behaviors of a single agent, and the multi-agent formation can be controlled through the combination of the basic behaviors only by researching the control method of each basic behavior. Basic behaviors generally include target tracking, obstacle avoidance, collision avoidance, formation generation, formation maintenance, and the like. The behavior-based method is controlled by mutual perception of all intelligent agents, the system is easy to realize distributed control and has good robustness, but the reliability of the system cannot be effectively ensured due to the fact that accurate mathematical model analysis cannot be combined.
The potential energy method is another formation control method different from the traditional method, all the steps of formation of the formation are combined together, the combined potential function is constructed, the control law of the intelligent agent is determined by the potential field function, other intelligent agents act on different potential forces of the intelligent agent, the expected formation graph is combined, the intelligent agent moves towards the expected formation graph, and when the intelligent agent respectively reaches the formation graph, the potential energy of the whole system is rather small. In the formation process of the formation, the formation target point is dynamically changed, which belongs to a dynamic formation method, and the intelligent agents reach a balanced state by continuously adjusting the distance between the intelligent agents; however, the rapidity and controllability of formation of the formation can not be grasped, because the formation of the formation is carried out according to the formation diagram of the relative distance, the relative positions of the intelligent bodies are actually determined in advance, so that the target positions of the intelligent bodies are probably not optimal, and because the formation diagram with fixed relative positions is kept on the whole formation target, if one intelligent body fails to reach the designated formation point, other intelligent bodies are always in a state of dynamically searching for the minimum potential energy point and move all the time.
Disclosure of Invention
The invention mainly solves the technical problem of providing a multi-agent formation method based on behavior selection, and solves the problem that the reliability and robustness of a system in the multi-agent formation process cannot be ensured.
In order to solve the technical problems, the invention adopts a technical scheme that: a behavior selection based multi-agent formation method is provided, which comprises the following steps:
step one, setting a detection profit calculation method related to the position:
(1) setting a target detection position profit region: the target detection position profit region setting specifically includes the following conditions: the income area is a plurality of fan-shaped areas taking the target detection center as a circle;
(2) the detection profit calculation method comprises the following steps: setting the detection point profit of the agent as a fixed profit related to the importance degree of the target point in a sector closer to the profit region; in a ring far away from a profit area, setting the profit of the detection point of the intelligent agent as a dynamic profit inversely proportional to the distance between the intelligent agent and a target point, and resetting the detection profit of the current operation cycle after the detection is finished;
step two, defining the specific behaviors of a plurality of agents: defining specific behaviors which may appear after behavior selection of the agent is completed according to the specific structure and function of the agent, limiting and setting and analyzing the excitation inhibition relation of each behavior, and if a certain agent behavior can be excited, verifying by adopting a basal ganglia channel model; if the stimulated agent behavior can be verified, executing the behavior;
determining the channel number of the basal ganglia, establishing a basal ganglia channel model, and initializing relevant parameters: the basal ganglia mathematical model comprises striatum, globus pallidus outer nucleus, subthalamic nucleus and globus pallidus inner nucleus;
(1) establishing a mathematical model of a behavior channel: each channel is represented by a leaky-integrate neuron:
Figure GDA0003403455430000041
wherein x is the state of the neuron, u is the input of the neuron, y is the output of the neuron, H is a step function, and the rest are model parameters;
yi C＝Si
integrating and processing cerebral cortex to obtain comprehensive importance index S combining all factors in the step oneiWhere i is the channel number, yi CCharacterizing the output of the ith channel in the cerebral cortex;
the state of striatum D1 may be described as:
Figure GDA0003403455430000042
wherein u isi SD1Is the neuronal input of the striatal D1, i channel, ai SD1Is the state of the channel neuron, yi SD1Is the output of the channel neuron, wCSD1Is the cerebral cortex to striatum D1 weight; 1+ lambda describes the stimulatory effect of dopamine on striatum D1,. epsilonSD1Is the output threshold of striatum D1;
the state of striatum D2 may be described as:
Figure GDA0003403455430000043
wherein u isi SD2Is the neuronal input of the striatal D2, i channel, ai SD2Is the state of the channel neuron, yi SD2Is the output of the channel neuron, wCSD2Is the weight of the cerebral cortex to the striatum D2, and 1-lambda describes the inhibitory effect of dopamine on striatum D2, epsilonSD2Is the output threshold of striatum D2;
the globus pallidus outer nucleus can be described as:
Figure GDA0003403455430000051
wherein u isi GPeIs the globus pallidus outer nucleus, neuron input of the i channel, ai GPeIs the state of the channel neuron, yi GPeIs the output of the channel neuron, wSD2GPeIs the weight, ε, from striatum D2 to the globus pallidus outer nucleusGPeIs the output threshold of the globus pallidus ectonucleus.
The model of the subthalamic nucleus can be described as:
Figure GDA0003403455430000052
wherein u isi STNIs the subthalamic nucleus, neuronal input of the i channel, ai STNIs the state of the channel neuron, yi STNIs the output of the channel neuron, wGPeSTNIs the weight of the globus pallidus ectonuclear to subthalamic nucleus, εSTNIs the output threshold of the subthalamic nucleus;
according to anatomical studies, the projection of the subthalamic nucleus to the globus pallidus kernel is quite diffuse, so when describing the mathematical model of the globus pallidus kernel, the output of the globus pallidus kernel is made to include the subthalamic nucleus inputs of the other left and right channels;
the description is as follows:
Figure GDA0003403455430000061
wherein u isi GPiIs the subthalamic nucleus, neuronal input of the i channel, ai GPiIs the state of the channel neuron, yi GPiIs the output of the channel neuron, wSD1GPeIs the weight of striatum D1 to the globus pallidus kernel, wSTNGPiThe output weight, ε, from the subthalamic nucleus to the globus pallidus nucleusGPiIs the output threshold of the globus pallidus kernel, and the value of w needs to be notedCSD1、wCSD2、wSTNGPiCharacterizing the excitatory connection, w, for positive valuesSD1GPi、wSD2GPe、wGPeSTNNegative values, characterize inhibitory connections;
and step four, correcting the channel model parameters of the basal ganglia, and correcting various parameters of the model in the step three when the overall operation condition of the intelligent agent deviates under the condition that the user preference is taken as a standard.
The invention has the beneficial effects that: the invention provides a multi-agent formation method based on behavior selection, which introduces a dynamic agent formation method based on behavior selection, effectively overcomes the defects of low robustness, low reliability and the like in the existing formation detection method, enables the technical scheme of the application to have wider universality by modifying channels through open parameters and various weights, introduces a target detection yield variable, can integrally improve the benefit of the multi-agent for detection tasks, and indirectly improves the system operation efficiency.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments are briefly introduced below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without inventive efforts, wherein:
FIG. 1 is a schematic diagram of the detection profit range in a preferred embodiment of a behavior selection-based multi-agent formation method of the present invention, in which reference numeral 3 is a detection target to be detected by an agent, reference numeral 2 is a fixed detection profit region, and reference numeral 1 is a dynamic profit region;
FIG. 2 is a schematic diagram of a 4-channel basal ganglia in a preferred embodiment of a behavior selection based multi-agent formation method of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Comprehensively considering the sum dis of the movement paths of the intelligent agent to the formation target positionsumSuitable moving speed v of intelligent bodyiThe agent adjusts the time T to the target point-oriented postureadjustTime T of movement of agent to target pointmoveThe detection yield of the agent arriving at the location after the next sampling period, the behavior selection, Tcost＝Tmove+Tadjust。
Referring to fig. 1-2, the embodiment of the invention includes the following steps:
step one, setting a detection profit calculation method related to the position:
(1) setting a target detection position profit region: the target detection position profit region is set according to the following conditions: the income area is a plurality of fan-shaped areas taking the target detection center as the circle center;
(2) the detection profit calculation method comprises the following steps: setting the detection point profit of the agent as a fixed profit related to the importance degree of the target point in a sector closer to the profit region; setting the detection point profit of the intelligent agent to be dynamic profit inversely proportional to the distance between the intelligent agent and a target point in a circular ring far away from a profit area, and resetting the detection profit of the current operation cycle after detection is finished;
Figure GDA0003403455430000081
wherein etaprofitFor dynamic benefits of the current target point, ηkFor the fixed yield of the current target point, r is the distance from the agent to the target detection point, r1For fixing the shortest distance between the detection profit region and the target detection point, r2The longest distance between the dynamic profit area and the target detection point is obtained;
step two, defining the concrete behaviors of the intelligent agent: defining specific behaviors which may appear after the behavior selection of the intelligent agent is completed according to the specific structure and the function of the intelligent agent; limiting and setting analysis is carried out on the excitation inhibition relation of each behavior, and if a certain intelligent agent behavior can be excited, a basal ganglia channel model is adopted for verification; if the stimulated agent behavior can be verified, executing the behavior;
the intelligent body posture behaviors are designed to have six types of actions of clockwise rotation, anticlockwise rotation, forward movement, backward movement, search and return:
Figure GDA0003403455430000091
wherein s isiDegree of importance, η, characterizing individual behaviorsi kIs a constant to be set;
determining the channel number of the basal ganglia, establishing a basal ganglia channel model and initializing relevant parameters;
the basal ganglia mathematical model comprises striatum, globus pallidus outer nucleus, subthalamic nucleus and globus pallidus inner nucleus;
(1) establishing a mathematical model of a behavior channel: each channel is represented by a leaky-integrate neuron:
Figure GDA0003403455430000092
where x is the state of the neuron, u is the input of the neuron, y is the output of the neuron, H is a step function, and the remainder are model parameters.
yi C＝Si
Integrating and processing cerebral cortex to obtain comprehensive importance index S combining all factors in the step oneiWhere i is the channel number, yi CCharacterizing the output of the ith channel in the cerebral cortex;
the state of striatum D1 may be described as:
Figure GDA0003403455430000101
wherein u isi SD1Is the neuronal input of the striatal D1, i channel, ai SD1Is the state of the channel neuron, yi SD1Is the output of the channel neuron, wCSD1Is the cerebral cortex to striatum D1 weight. 1+ lambda describes the stimulatory effect of dopamine on striatum D1,. epsilonSD1Is the output threshold of striatum D1;
the state of striatum D2 may be described as:
Figure GDA0003403455430000102
wherein u isi SD2Is the neuronal input of the striatal D2, i channel, ai SD2Is the state of the channel neuron, yi SD2Is the output of the channel neuron, wCSD2Is the weight of the cerebral cortex to the striatum D2, and 1-lambda describes the inhibitory effect of dopamine on striatum D2, epsilonSD2Is the output threshold of striatum D2;
the globus pallidus outer nucleus can be described as:
Figure GDA0003403455430000111
wherein u isi GPeIs a Chinese characterOuter nucleus of the white sphere, neuronal input of the i channel, ai GPeIs the state of the channel neuron, yi GPeIs the output of the channel neuron, wSD2GPeIs the weight, ε, from striatum D2 to the globus pallidus outer nucleusGPeIs the output threshold of the globus pallidus ectonucleus;
the model of the subthalamic nucleus can be described as:
Figure GDA0003403455430000112
wherein u isi STNIs the subthalamic nucleus, neuronal input of the i channel, ai STNIs the state of the channel neuron, yi STNIs the output of the channel neuron, wGPeSTNIs the weight of the globus pallidus ectonuclear to subthalamic nucleus, εSTNIs the output threshold of the subthalamic nucleus;
according to anatomical studies, the projection of the subthalamic nucleus to the globus pallidus kernel is quite diffuse, so when describing the mathematical model of the globus pallidus kernel, the output of the globus pallidus kernel is made to include the subthalamic nucleus inputs of other left and right channels;
the description is as follows:
Figure GDA0003403455430000121
wherein u isi GPiIs the subthalamic nucleus, neuronal input of the i channel, ai GPiIs the state of the channel neuron, yi GPiIs the output of the channel neuron, wSD1GPeIs the weight of striatum D1 to the globus pallidus kernel, wSTNGPiThe output weight, ε, from the subthalamic nucleus to the globus pallidus nucleusGPiIs the output threshold of the globus pallidus kernel, and the value of w needs to be notedCSD1、wCSD2、wSTNGPiCharacterizing the excitatory connection, w, for positive valuesSD1GPi、wSD2GPe、wGPeSTNNegative values, characterize inhibitory connections;
step four: correcting channel model parameters of the basal ganglia: and when the overall operation condition of the intelligent agent deviates under the condition of taking the user preference as a standard, correcting each parameter of the model in the step three. And finally, selecting the behavior of the intelligent agent according to the globus pallidus kernel output.
In summary, the behavior selection-based multi-agent formation method disclosed by the invention enhances the reliability and robustness of the system by introducing the basal ganglia behavior selection formation strategy, and can be applied to agent clusters with different functions, different quantities and scales, different detection targets, different motion modes and target detection modes.
The above description is only an embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by the present specification, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.

Claims (4)

1. A multi-agent formation method based on behavior selection, comprising the steps of:
setting a position-related detection benefit calculation method, wherein a benefit area is a plurality of fan-shaped areas with a target detection center as a circle center, and in a fan shape close to the benefit area, setting detection point benefits of an intelligent agent as fixed benefits related to the importance degree of a target point; setting detection point benefits of the intelligent agent to be dynamic benefits which are inversely proportional to the distance between the intelligent agent and a target point in a circular ring far away from a benefit area, and resetting the detection benefits of the current operation period after detection is finished;
step two, defining the specific behaviors of a plurality of agents: defining specific behaviors which may appear after the behavior selection of the intelligent agent is completed according to the specific structure and the function of the intelligent agent; limiting and setting analysis is carried out on the excitation inhibition relation of each behavior, and if a certain intelligent agent behavior can be excited, a basal ganglia channel model is adopted for verification; if the stimulated agent behavior can be verified, executing the behavior; designing the posture behaviors of the intelligent body to have six types of clockwise rotation, anticlockwise rotation, forward movement, backward movement, search and return;
determining the channel number of the basal ganglia, establishing a basal ganglia channel model and initializing relevant parameters;
and step four, correcting the channel model parameters of the basal ganglia.
2. The behavioral selection-based multi-agent formation method according to claim 1, wherein in step three, the basal ganglia mathematical model includes striatum, globus pallidus ectonucleus, subthalamic nucleus and globus pallidus nucleus.
3. The behavior selection based multi-agent queuing method according to claim 1, wherein in step four, when the overall operation condition of the agents deviates under the condition of taking user preference as standard, the correction of each parameter of the model in step three is performed.
4. The behavior selection based multi-agent formation method according to claim 1, wherein a plurality of agents are allowed to have different functional characteristics, and the overall work efficiency is improved by behavior selection control.
CN201810286164.3A 2018-04-03 2018-04-03 Multi-agent formation method based on behavior selection Active CN108646550B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810286164.3A CN108646550B (en) 2018-04-03 2018-04-03 Multi-agent formation method based on behavior selection

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810286164.3A CN108646550B (en) 2018-04-03 2018-04-03 Multi-agent formation method based on behavior selection

Publications (2)

Publication Number Publication Date
CN108646550A CN108646550A (en) 2018-10-12
CN108646550B true CN108646550B (en) 2022-03-22

Family

ID=63745435

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810286164.3A Active CN108646550B (en) 2018-04-03 2018-04-03 Multi-agent formation method based on behavior selection

Country Status (1)

Country Link
CN (1) CN108646550B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108897315A (en) * 2018-06-12 2018-11-27 南京理工大学 A kind of Multi-agent Team Formation
CN109725532B (en) * 2018-12-24 2021-11-16 杭州电子科技大学 Relative distance control and self-adaptive correction method applied to multiple intelligent agents
CN113268893B (en) * 2021-07-19 2021-10-01 中国科学院自动化研究所 Group trapping method and device based on communication maintenance constraint

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8379967B1 (en) * 2008-12-04 2013-02-19 Lockheed Martin Corporation Swarm control systems and methods
CN105205533A (en) * 2015-09-29 2015-12-30 华北理工大学 Development automatic machine with brain cognition mechanism and learning method of development automatic machine
CN105467981A (en) * 2015-12-24 2016-04-06 中南大学 Method and device for formation aiming at multiple intelligent agents
JP2016095754A (en) * 2014-11-17 2016-05-26 日本電信電話株式会社 Action control system, and method, and program, for the same
CN106843269A (en) * 2017-01-22 2017-06-13 北京航空航天大学 A kind of unmanned plane formation method based on small birds cluster fly mechanics
CN107121986A (en) * 2017-05-24 2017-09-01 浙江大学 The method that a kind of unmanned plane flight pattern of Behavior-based control is kept
CN109144047A (en) * 2018-06-12 2019-01-04 南京理工大学 A kind of finite time formation control method based on multi-agent system

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104925059B (en) * 2015-06-02 2017-07-11 奇瑞汽车股份有限公司 Automatically with car method and device

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8379967B1 (en) * 2008-12-04 2013-02-19 Lockheed Martin Corporation Swarm control systems and methods
JP2016095754A (en) * 2014-11-17 2016-05-26 日本電信電話株式会社 Action control system, and method, and program, for the same
CN105205533A (en) * 2015-09-29 2015-12-30 华北理工大学 Development automatic machine with brain cognition mechanism and learning method of development automatic machine
CN105467981A (en) * 2015-12-24 2016-04-06 中南大学 Method and device for formation aiming at multiple intelligent agents
CN106843269A (en) * 2017-01-22 2017-06-13 北京航空航天大学 A kind of unmanned plane formation method based on small birds cluster fly mechanics
CN107121986A (en) * 2017-05-24 2017-09-01 浙江大学 The method that a kind of unmanned plane flight pattern of Behavior-based control is kept
CN109144047A (en) * 2018-06-12 2019-01-04 南京理工大学 A kind of finite time formation control method based on multi-agent system

Non-Patent Citations (6)

* Cited by examiner, † Cited by third party
Title
Behavior-based Formation Control for Multi-robot Teams;Tucker Balch 等;《IEEE》;19981231;全文 *
基于基底神经节的机器人行为选择与行为序列学习方法;王义萍;《中国博士学位论文全文数据库 信息科技辑》;20120115;第5,11,21-27页及图3.1 *
基底神经节的尖峰神经元网络模型及其在机器人中的应用;王义萍 等;《南京理工大学学报（自然科学版）》;20101231;第34卷(第6期);全文 *
机器人行为选择机制综述;王义萍 等;《机器人》;20091231;第31卷(第5期);全文 *
水下机器人编队***研究;王兢;《中国优秀博硕士学位论文全文数据库 (硕士) 信息科技辑》;20031215;第13-15页 *
第六章 多智能体及行为;吴建国编著;《仿生机器人及其智能运动控制》;20050228;第1-10页 *

Also Published As

Publication number Publication date
CN108646550A (en) 2018-10-12

Similar Documents

Publication Publication Date Title
WO2021238191A1 (en) Positioning compensation method and apparatus for robot
CN108646550B (en) Multi-agent formation method based on behavior selection
Bode Neural networks for cost estimation
Ren et al. A scatter learning particle swarm optimization algorithm for multimodal problems
CN105554873B (en) A kind of Wireless Sensor Network Located Algorithm based on PSO-GA-RBF-HOP
CN107885232A (en) A kind of filtering method for how tactful maneuver tracking
CN110471289B (en) Adaptive path tracking method and system based on visual navigation mobile equipment
CN105427241B (en) Distortion correction method for large-view-field display equipment
CN108748160A (en) Manipulator motion planning method based on particle cluster algorithm on multiple populations
CN108897315A (en) A kind of Multi-agent Team Formation
CN110490422A (en) A kind of target fighting efficiency method for situation assessment based on game cloud model
CN115420294A (en) Unmanned aerial vehicle path planning method and system based on improved artificial bee colony algorithm
Zernetsch et al. Trajectory forecasts with uncertainties of vulnerable road users by means of neural networks
CN109799703A (en) A kind of population Auto-disturbance-rejection Control, device and storage medium
CN110162077A (en) A kind of unmanned aerial vehicle flight path planing method based on flying fish algorithm
Ma et al. Multi-AUV collaborative operation based on time-varying navigation map and dynamic grid model
US20230166397A1 (en) Method for obstacle avoidance in degraded environments of robots based on intrinsic plasticity of snn
CN111002302A (en) Mechanical arm grabbing track planning method combining Gaussian mixture model and dynamic system
CN113281993A (en) Greedy K-mean self-organizing neural network multi-robot path planning method
Liu A novel path planning method for aerial UAV based on improved genetic algorithm
Wang et al. Deep learning based missile trajectory prediction
CN114670856B (en) Parameter self-tuning longitudinal control method and system based on BP neural network
CN110145966A (en) A kind of finite time convergence control Global robust sliding formwork method of guidance
CN116149363A (en) Energy consumption optimization method for formation transformation of multi-unmanned aerial vehicle system
Bataineh et al. Artificial neural network-based prediction of human posture

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