Development of Co-operative Distributed Outdoor Type Lawn Mowing System by Dynamical Task Allocation of Small-Size Robots

This project is to develop a lawn-mowing system by a group of small, cheap, and not so much precisely designed robots. The purpose is to realize reliable, efficient, and low-cost task performing system for a wide病rea outdoor task. In this research, we developed an autonomic charging system, which consists of a charging house with automatic shutter and an infrared driven homing system. This is the basis for giving autonomy for a group of robots. To allocate appropriate tasks to each robot, we developed an on-line task allocation system that performs allocation during communication by a dynamic communication channel connected by ad-hoc network. The design and tuning of the algorithm is based on an agent platform simulator by OAA architecture. The on-line connection and communication in ad-hoc network is realized by a transport protocol based on an SOAP message, which realizes rich communication involving status of machines and even for communication. Finally, we could develop an automatic fail-recovery algorithm, which is important issue of the outdoor robot designed at low cost and imprecise. The method is based on a trial-and-error recovery, with the acquisition of experience for successful escape, which then allows the robot to find an efficient way to escape for a similar situation of stuck.

This project is to develop a lawn-mowing system by a group of small, cheap, and not so much precisely designed robots. The purpose is to realize reliable, efficient, and low-cost task performing system for a wide病rea outdoor task. In this research, we developed an autonomic charging system, which consists of a charging house with automatic shutter and an infrared driven homing system. This is the basis for giving autonomy for a group of robots. To allocate appropriate tasks to each robot, we developed an on-line task allocation system that performs allocation during communication by a dynamic communication channel connected by ad-hoc network. The design and tuning of the algorithm is based on an agent platform simulator by OAA architecture. The on-line connection and communication in ad-hoc network is realized by a transport protocol based on an SOAP message, which realizes rich communication involving status of machines and even for communication. Finally, we could develop an automatic fail-recovery algorithm, which is important issue of the outdoor robot designed at low cost and imprecise. The method is based on a trial-and-error recovery, with the acquisition of experience for successful escape, which then allows the robot to find an efficient way to escape for a similar situation of stuck.