Multi-agent Robotic Systems Research Articles

A Collision Prevention Technique for Multi-Agent Robot Systems with Topology Map

A Collision Prevention Technique for Multi-Agent Robot Systems with Topology Map

UB SWARM: HARDWARE IMPLEMENTATION OF HETEROGENEOUS SWARM ROBOT WITH FAULT DETECTION AND POWER MANAGEMENT

In this work we present the hardware architecture of a mobile heterogeneous robot swarm, designed and implemented at the Interdisciplinary Robotics, Intelligent Sensing and Control (RISC) Laboratory, University of Bridgeport. Most of the recent advances in swarm robotics have mainly focused on homogeneous robot swarms and their applications. Developing and coordinating a multi-agent robot system with heterogeneity and a larger behavioral repertoire is a great challenge. To give swarm hardware heterogeneity we have equipped each swarm robot with different set of sensors, actuators, control and communication units, power supply, and an interconnection mechanism. This paper discusses the hardware heterogeneity of the robotic swarm and its challenges. Another issue addressed in paper is the active power management of the robotic agents. The power consumption of each robot in the UB robot swarm is calculated and the power management technique is also explained in this paper. We applied this heterogeneous robot swarm to perform three sample tasks – Mapping task, human rescue task and wall painting task.

Man-Machine Interface for Autonomous Robots and Multi-Agent Robotic Systems

The article provides an overview of tools and methods of man-machine interface for special types of robotic systems. It continues the series of previous publications on multi-robot systems that can provide a cooperative solution to the specified task. Although most multi-robot systems research is centered around the problems of group control, it is obvious that the effectiveness of such systems will be determined by the capabilities of man-machine interface designed to interact with them. In this article we show the evolution of man-machine interfaces starting from manually controlled robots and finishing with autonomous robots. We also discuss a variety of approaches to implementation of such interfaces. From the analysis of modern interfaces functional capabilities we offer the construction principles of man-machine interface for multi-agent robotic systems. These principles are based on complex automatization of information preprocessing for a multi-robot systems at every stage of their operation; intellectualization of decision-making process; software modularity and unification. The structure and composition of software and algorithmic tools for multi-robot group's task description are justified including the systems responsible for information input/output; task scenario model formation; agents number estimation; task planning; multi-robot system modelling; information gathering and interpretation. Supervising software is presented that controls the coordinated functioning of man-machine interface subsystems. The results of the simulation are given including the task scenario formation and its execution monitoring.

Global swarming while preserving connectivity via Lagrange–Poincarè equations

In this paper, we exploit symmetry properties of multi-agent robot systems to design control laws that preserve connectivity while swarming. We start by designing a connectivity control law for agents with configuration spaces R3 and SO(3) that is invariant under the action of the special Euclidean group SE(3) and the special orthogonal group SO(3), respectively. Therefore, the dynamics of such multi-agent systems is amenable to be reduced by these group actions. We then utilize the Lagrange–Poincaré equations that split the Euler–Lagrange equations for the multi-agent system into horizontal and vertical parts. The invariance of the connectivity controller implies that its control effort has zero vertical component. We then use the resulting vertical equations of motion to design a control law that asymptotically stabilizes the centroid and the orientation of the swarm at a desired pose.

Multi-Agent Robot System and Events Executing Using GPS Cloud

Large number of autonomous robot solutions exists for various missions and domains. These robots are sufficient for the missions they are built for. At the same time each of them has limited functional and physical capabilities. Multi agent systems can be used to remove these limits. However it is true only in case when the system ensures effective interaction among the robots i.e. enables their social behavior. Usually it is hard to implement such capabilities directly into robots due to functional and physical limitations and heterogeneity of the team. One of possible solutions is to implement a behavior sensors management for the robots. It should collect events, allocate subtasks to specific robots and monitor the execution of the assigned tasks. In order to avoid inherent drawback of fully centralized systems a significant level of autonomy has to be preserved. Intelligent agents fulfill these requirements. Therefore we propose a multi-agent system’s architecture for safe road application with GPS tool. It can be used to control the car speed and to adjust it in case of danger.

Multi-Agent Robot System and Events Executing Using GPS Cloud

Large number of autonomous robot solutions exists for various missions and domains. These robots are sufficient for the missions they are built for. At the same time each of them has limited functional and physical capabilities. Multi agent systems can be used to remove these limits. However it is true only in case when the system ensures effective interaction among the robots i.e. enables their social behavior. Usually it is hard to implement such capabilities directly into robots due to functional and physical limitations and heterogeneity of the team. One of possible solutions is to implement a behavior sensors management for the robots. [1] It should collect events, allocate subtasks to specific robots and monitor the execution of the assigned tasks. In order to avoid inherent drawback of fully centralized systems a significant level of autonomy has to be preserved. Intelligent agents fulfill these requirements. Therefore we propose a multi-agent system’s architecture for safe road application with GPS tool. It can be used to control the car speed and to adjust it in case of danger.

Multi-Agent Robot System and Events Executing Using GPS Cloud

Multi-Agent Robot System and Events Executing Using GPS Cloud

Implementation of the Communication Network for the Multi-Agent Robotic Systems

The paper proposes a structure of self-organizing communication networks of the multi-robot system. The considered networks are organized on the basis of two radio channels: to transmit control (telemetry data) and to transmit data (video flow), which have the higher bandwidth requirements. The algorithm of self-organizing network is designed and presented based on the wideband channel quality control: data of the received signal strength indicator (RSSI), the capacity and the packet error ratio (PER).

Учебно-исследовательский комплекс на базе LEGO MINDSTORMS NXT 2.0 для отработки технологий многоагентных робототехнических систем

In today's robotics, an increasing attention is devoted to Multi-Agent Robotic Systems (MARS), which allow us to accomplish a wide range of tasks with high efficiency due to distribution of the tasks among several agents. The present paper is concerned with the development of an educational research complex intended to improve the algorithms and methods of robot group control at the Chair of Management Problems of the Moscow State University of Computer Science, Radio Engineering and Electronics (MIREA). As a result of an analytical review of the possible approaches to the development of the educational research complex within the framework of the present project, it was suggested to use Lego Mindstorms NXT 2.0 kit in order to save time and financial resources. Despite the simplicity of the structure due to the scantiness of the Lego set of the basic elements, the developed robot, however, provides the minimum of the necessary functional possibilities for solving of different tasks to work out the methods and algorithms of the robot group control. Planning of the appropriate actions and behavior of the robots is based on the technology of the frame-based structures for providing models (scenarios) of the typical behavior and the subsequent selection of a particular model, depending on the situation. Determination of the agents' current location and their orientation is carried out by an external vision system based on the optical character recognition (optical glyphs) technology. For the wireless data exchange between the control centre and the agents, Bluetooth technology of data transmission is used. The planning and distribution of tasks by the control centre of the MARS is realized in the form of software written in C#. The man-machine interface developed within the present research has all the necessary modules allowing us to realize the problem statement and to efficiently control its implementation. As an application example, we consider the task of building block constructions. The construction blocks are represented by cubes of different colours to be installed in a given sequence, corresponding to the construction process. The results of the experiments fulfilled in order to test our hypotheses and construction principles of the MARS are presented.

Механизмы интеллектуальных обратных связей, обработки знаний и самообучения в системах управления автономными роботами и мультиагентными робототехническими группировками

The article is devoted to analysis of various approaches aimed to ensure adaptive properties of the automatic control systems designed for maintenance of the demanded reliability and functioning quality in the presence of disturbances and uncertainty factors. A review is provided of the development of the creation principles for the automatic control systems - from organization of an internal feedback, up to creation of intelligent feedback contours with self-training mechanisms. Examples of various robots equipped with intelligent self-learning control systems are presented. A list of tasks of self-learning of the intelligent control systems is presented both for the autonomous robots and for the multi-agent robot systems. A generalized structure of a control system of the autonomous mobile robots is offered based on the intelligent feedback as means of new knowledge formation. Knowledge formation is possible in a self-learning mode on the basis of processing of the accumulated sensory information. The main problems with creation of the intelligent self-learning control systems are discussed. It is shown that one of the most perspective approaches to realization of the self-learning process in the intelligent control systems of the autonomous robots and multi-robot systems is related to the method of classification of tree formation. The article gives an example of a practical solution to the self-learning problem of the autonomous mobile robots. The solution is based on the automatic knowledge formation of the passability characteristics of the surface underlying a robot. The results of a complex modeling are presented, which testify to the expediency and efficiency of incorporation of the self-learning methods in the structure of a mobile robot intelligent onboard control system for improvement of its adaptation properties. The prospects of the self-learning processes are described for the autonomous robots operating in the structure of the multi-agent systems. As a prove experimental estimates of the efficiency are given for application of the self-learning methods aimed at adaptation of the autonomous mobile robots under conditions of environmental uncertainty on the basis of a mutual exchange of the terrain passability knowledge.

A School of Robotic Fish for Mariculture Monitoring in the Sea Coast

This paper presents a multi-agent robotic fish system used for mariculture monitoring. Autonomous robotic fish is designed to swim underwater to collect marine information such as water temperature and pollution level. Each robotic fish has 5 degrees of freedom for controlling its depth and speed by mimicking a sea carp. Its bionic body design enables it to have high swimming efficiency and less disturbance to the surrounding sea lives. Several onboard sensors are equipped for autonomous 3D navigation tasks such as path planning, obstacle avoidance and depth maintenance. A robotic buoy floating on the water surface is deployed as a control hub to communicate with individual robots, which in turn form a multi-agent system to monitor and cover a large scale sea coast cooperatively. Both laboratory experiments and field testing have been conducted to verify the feasibility and performance of the proposed multi-agent system.

2A1-L05 強化学習におけるクラウドの活用を前提とした知識統合手法の開発

This paper describes an assimilation method of knowledge obtained by reinforcement learning agents. In recent years, multi-agent robot system (MARS) utilizing reinforcement learning and its knowledge sharing methodology have been deployed in real-world situations. For example, knowledge assimilation methods such as simple-sum are proposed for merging the knowledge obtained from distributed learning agents in different tasks. Those methodologies, however, have not been fully discussed systematically. Furthermore, existing researches have only been dealing with the knowledge which is obtained from individual tasks. To realize a knowledge assimilation method with various knowledge of agents in multi-agent reinforcement learning systems, we focused on a cloud computing resources. In this paper, we propose a knowledge assimilation method utilizing a cloud computing resource, and we confirmed the effectiveness of knowledge assimilation method in a shortest path problem.

2A1-L06 ヘテロジーニアス間転移学習のための知識再利用法の検討

This paper describes an influence of heterogeneity for a transfer learning in heterogeneous multiple robots. A multi-agent robot system using reinforcement learning (Multi-agent reinforcement learning: MARL) is effective methodology to obtain efficient behaviors autonomously in dynamic environment. However, the MARL has a problem such as prolongation in learning time. The existing research indicated effectiveness of transfer learning for reducing of learning time in reinforcement learning. Transfer learning is a framework of reuse of knowledge which is obtained by reinforcement learning. In our prior research, we proposed and applied the parameter, transfer rate, to the transfer learning method for heterogeneous agents. The parameter adjusts the degree of reuse of past knowledge toward a newly obtained knowledge in a new task. However, transferability is depending on heterogeneity factors such as difficulty of tasks and agents' functions. In this paper, we investigate the effectiveness of transfer rate with heterogeneous agents and environments by conducting a computer simulation.

Distributed containment control and state observer design for multi-agent robotic system

Aiming at the distributed containment control problem for multi-agent robotic system with stationary leaders, a position-sensorless control strategy is proposed based on binocular visual servo and feedback dissipative Hamilton theory. Firstly, the topological relationship among the robots is built by using directed graph, which makes each follower determine its neighbour robots. Secondly, a binocular visual servo model is established, so as to describe the dynamic information of the tracking target point for each follower through the neighbour robots' image information. Thirdly, the distributed containment controllers are designed based on the pre-feedback control strategy and feedback dissipative Hamilton theory; furthermore, in the case that each follower is not equipped with internal position sensor, the full-order state observers are developed by using the 'expansion + feedback' method, and then realise the position-sensorless control. Finally, a simulation is carried out on multiple two degrees-of-freedom (DOF) robots, and results show the validity of the proposed control strategy.

A hierarchical CSP search for path planning of cooperating self-reconfigurable mobile fixtures

The paper presents the application of artificial intelligence tools for the path planning of complex multi-agent robotic systems. In particular, a solution is proposed to the planning problem for the conjoint operation of two or more mobile robotic fixtures used for the manufacturing of large workpieces, like those used in the aerospace industry. Such fixturing systems have been recently designed and tested, raising hopes to better satisfy the dynamic conditions of modern manufacturing, with its increasing emphasis on flexibility, adaptability, and automation. The proposed planning method is novel in two fundamental aspects. First, it interprets planning as a constraint satisfaction problem (CSP), rather than as a constrained optimisation, an approach ubiquitous in the path and motion planning literature. Secondly, the formulated CSP is solved by a hierarchy of incremental state space search algorithms which differ in some way from the existing state of the art. This hierarchy includes levels related to the robot and workpiece arrangement parameters and to three components of mobile fixture agents: a supporting head, a mobile base, and a parallel manipulator, respectively. Due to the use of CSP search, the planner constitutes a largely application-independent framework, on the basis of which specific industrial implementations can be defined by supplying the relevant physical, geometrical, and time-related constraints.

Transfer Learning Method Using Ontology for Heterogeneous Multi-agent Reinforcement Learning

This paper presents a framework, called the knowledge co-creation framework (KCF), for heterogeneous multiagent robot systems that use a transfer learning method. A multiagent robot system (MARS) that utilizes reinforcement learning and a transfer learning method has recently been studied in realworld situations. In MARS, autonomous agents obtain behavior autonomously through multi-agent reinforcement learning and the transfer learning method enables the reuse of the knowledge of other robots’ behavior, such as for cooperative behavior. Those methods, however, have not been fully and systematically discussed. To address this, KCF leverages the transfer learning method and cloud-computing resources. In prior research, we developed ontology-based inter-task mapping as a core technology for hierarchical transfer learning (HTL) method and investigated its effectiveness in a dynamic multi-agent environment. The HTL method hierarchically abstracts obtained knowledge by ontological methods. Here, we evaluate the effectiveness of HTL with a basic experimental setup that considers two types of ontology: action and state.

Object Tracking with a Multiagent Robot System and a Stereo Vision Camera

When working with a robot in terms of object manipulation the essential information is relative position between robot's tool center point (TCP) and the object of interest. This paper proposes a method of frame relative displacement and describes a working multiagent robot application that can be used for tracking, tooling or handling operations with the use of stereo vision in unstructured laboratory environment. Robot system is composed of two Fanuc robot arms, one of which carries a stereo vision camera system and the other which is guided in relation to object of interest. The latter robot has a marker that is used for navigation between the robot and the object of interest. Image processing, marker detection, 3-D coordinates extraction, coordinate system transformations, offset coordinates calculation and communication are handled using c++ multithread program and TCP/IP protocol.

Rigid Folding in Robotic Multi-agent Systems

The paper seeks to investigate the potential of biologically inspired multi-agent systems for architectural applications. The generation of ‘form’ through the process of ‘rigid folding’ represents a complex architectural procedure, which we seek to perform by employing a number of simple collaborative agents. Through a prototypic approach, closely matched with the constraints of a real world application, we want to demonstrate that the central architectural needs of form adaptability and resilience can be increased with this method. The established prototype is intended to serve as the ‘embodiment’ for different codes and strategies to test. Its shape transitions are based on the principal of folding, in which the triangle mesh topology serves as the basis for its folding patterns.

Global Swarming while Preserving Connectivity via Lagrange-Poincarè Equations

In this paper, we exploit symmetry properties of multi-agent robot systems to design control laws that preserve connectivity while swarming. We start by showing that the connectivity controller is invariant under the action of the special Euclidean group SE(3) and therefore is amenable to reduction of the dynamics by this action. We then utilize the reduced Euler-Lagrange equations that split the Euler-Lagrange equations for the multi-agent system into horizontal and vertical parts. The invariance of the connectivity controller implies that its control effort has zero vertical component. We then use the resulting vertical equations of motion to design a control law that asymptotically stabilizes the centroid and the orientation of the swarm at a desired pose.

Multi-agent compositional stability exploiting system symmetries

This paper considers nonlinear symmetric control systems. By exploiting the symmetric structure of the system, stability results are derived that are independent of the number of components in the system. This work contributes to the fields of research directed toward compositionality and composability of large-scale system in that a system can be “built-up” by adding components while maintaining system stability. The modeling framework developed in this paper is a generalization of many existing results which focus on interconnected systems with specific dynamics. The main utility of the stability result is one of scalability or compositionality. If the system is stable for a given number of components, under appropriate conditions stability is then guaranteed for a larger system composed of the same type of components which are interconnected in a manner consistent with the smaller system. The results are general and applicable to a wide class of problems. The examples in this paper focus on the formation control problems for multi-agent robotic systems.