Distributed Energy-Efficient Mobile Robots

Purdue University, West Lafayette IN

Investigators

Abstract

Mobile robots (mobots) have a wide range of applications in humanitarian operations, including search and rescue operations. To successfully accomplish these missions, mobots must maintain high mobility by carrying their own energy sources, and they must make efficient use of their limited energy resources in order to be capable of operating for extended periods of time. In this project, the PI and his team will investigate a new methodology of constructing distributed energy-efficient mobots. They will design, implement, and evaluate a team of mobots controlled by using biologically inspired engineering principles, and where inter-mobot communication is conducted through wireless ad hoc networks with peer-to-peer overlay protocols. Speci?cally, the team will study: (1) Biologically inspired algorithms for sensing, planning, coordination, and control. This is motivated by the observation that many creatures, through natural selection, use energy efficiently. Soft computing and multi-agent systems will be adopted to manage the team coordination as well as the activities of individual mobots. Neuro-fuzzy logic will be used to control multiple robots with energy efficiency. (2) Energy-efficient, scalable, and robust wireless networks for inter-mobot communication. Peer-to-peer overlay networks will be constructed on mobile ad hoc networks to support the high mobility and to handle the possibility of losing or adding mobots during missions. Packets will be transmitted through energy-efficient routes; these routes will be dynamically determined based on the current locations and the remaining energy of the mobots. (3) Energy models for each component in a mobot, the interactions among the components inside each mobot, and the interactions among mobots. These models will be used to evaluate the energy efficiency of the control algorithms developed in (1) and the communication mechanism presented in (2). In addition, a comprehensive simulator will be constructed to study a large number of mobots: the team movement, con?guration, coordination, and the communication among them. The simulator will also validate whether biological systems indeed use energy efficiently. The outcomes of this research will consist of: (i) biologically inspired neuro-fuzzy control algorithms to improve energy efficiency; (ii) efficient inter-mobot communication mechanisms; (iii) three mobot prototypes for experiments and instrumentation; and (iv) a detailed simulator to study various scenarios with a large team of mobots. Broader Impacts: This project has broader impacts on society, education, and outreach. Mobots will be able to operate for signi?cantly longer times in humanitarian missions and outer space exploration. Research results will be incorporated into graduate and undergraduate courses at Purdue, and the lab will host an open session to local K-12 students every year to demonstrate the mobot prototypes and to encourage participation by students in engineering and science activities.