The research objective of this Faculty Early Career Development (CAREER) project is to develop a new systematic framework for the synthesis and analysis of distributed learning and cooperative control algorithms for multi-agent systems with applications in environmental sciences. Practical algorithms will be developed and analyzed so that a resource-constrained, multi-agent system can collect spatial measurements of a scalar field in order to (i) self-calibrate the environmental model, (ii) recursively learn the unknown field based on the calibrated model and (iii) perform tasks such as the exploration, estimation, prediction and maximum seeking of a scalar field in a robustly intelligent manner. Theoretically sound, implementable control laws will be developed and analyzed using concepts such as stochastic approximation theory and kernel regression. The conventional inverse problem approach based on physical transport models is too computationally costly for resource-constrained, multi-agent systems. In contrast, emphasizing practicality and usefulness, this research relies extensively on phenomenological and statistical modeling techniques such as Gaussian processes and kernel regression to represent fields undergoing transport phenomena. Outcomes will be tested on mobile robot platforms under realistic transport phenomena.
If successful, the results of this research will have a significant impact on environmental sciences. The proposed multi-agent systems, combined with different types of chemical sensors, will have important applications such as the prediction and tracing of algal blooms and of toxic chemicals in real-life situations. This interdisciplinary project will offer training experiences for graduate and undergraduate students and provide opportunities to foster collaboration with departments of Civil and Environmental Engineering and Fisheries and Wildlife. In collaboration with the Diversity Programs Office and the Office of Recruitment and K-12 Outreach at Michigan State University, the educational program will offer interactive seminars on biologically inspired mobile robots to K-12 and underrepresented students in summer residential programs.
