Computer Sciences and Information Technology
Title : | Cooperative Control of Networked Multi-agent Systems utilizing Negative Imaginary Theory with an Application to Multi-robot Systems |
Area of research : | Computer Sciences and Information Technology |
Principal Investigator : | Dr. Parijat Bhowmick, Indian Institute Of Technology (IIT) Guwahati, Assam |
Timeline Start Year : | 2022 |
Timeline End Year : | 2024 |
Contact info : | parijat.bhowmick@iitg.ac.in |
Equipments : | Miniature two-wheeled mobile robots
Overhead Camera
High-end Bots
Work Station |
Details
Executive Summary : | Cooperative Control of Multi-Robot Systems (MRS) is a prominent field of research that has witnessed immense growth in the last fifteen years, mainly due to its potential applications in solving real-world problems and complex repetitive tasks. Examples include search-and-rescue operations in hazardous environments, security and surveillance, precision agriculture, factory automation, surgical robotics, etc. Some of these tasks can be solved as a cooperative control problem by robotic agents. This project aims to utilize Negative Imaginary (NI) Systems theory to design state-of-the-art control algorithms, tools and methodologies for safe, reliable and efficient operation of networked MRS in a Multi-agent Systems (MAS) framework. NI theory was introduced in 2007-08 and was inspired by the 'Positive Position Feedback Control' of lightly-damped mechanical systems. NI theory is analogous to passivity and offers a standalone robust control analysis and synthesis framework. The acronym NI has stemmed from the fact that such systems exhibit a 'negative imaginary' frequency response, i.e., the imaginary part of an NI transfer function takes negative values for all positive frequencies. The primary motivation for applying NI theory in this project is that many MAS (including a class of MRS) can be modelled as (transformed into) a group of single/double integrator agents, which inherently satisfy NI property. This project will first study the behaviour of a class of MAS (including MRS) and finish the modelling part relying on the NI dynamics. In the second step, the project will design a suitable network topology (directed/undirected) to capture the interactions among the agents. The project will then develop a novel cooperative control scheme utilizing NI properties, which consists of three main parts: a game theory-based clustering algorithm, a distributed group formation tracking scheme and a distributed containment control scheme. The clustering algorithm distributes the agents into several subgroups based on the positions, priorities and complexity of the assigned targets. The formation tracking scheme drives the leaders to achieve the desired formations surrounding the targets and by the containment control scheme, the follower agents are navigated towards a safe zone guarded by the leaders. At the fourth step, all three parts will be integrated and a software package will be created to easily implement the scheme in practice. Then, exhaustive simulation (using Matlab/Gazebo/Webot software) case studies will be performed, which will apply the scheme on several MRS to test its feasibility and usefulness. The project will also investigate the obstacle and collision avoidance capability of the scheme and its robustness to sudden loss of agents, communication delays and model uncertainties. Finally, a low-cost, lab-based multi-robot test rig will be developed to validate the scheme on a fleet of miniature two-wheeled mobile robots and sophisticated bots. |
Total Budget (INR): | 9,40,500 |
Organizations involved