A multi-agent based robot telesupervision architecture for hazardous materials detection

This paper focuses on the development of a multi-agent based robot telesupervision architecture for hazardous materials detection. By combining the supervisors' advanced reasoning capabilities with the robots' local autonomous capacities, this architecture is designed to maximize the safety and efficiency of robot supervisors to detect and deal with hazardous materials. The multi-agent framework not only supports instantiation of predefined abstract agents to meet a specific task, but also allows new abstract agents to be integrated. The key feature is that it is sufficiently flexible to support a wide variety of sensors, attachments, task-oriented detection instruments, etc., with regard to physical integration and to support a collaborative and dynamically changing bi-directional human-robot interaction rather than unidirectional human command. We have been using the robot telesupervision architecture in the application of hazardous materials detection for the past several years. Experimental results show it is applicable, adaptable and extendable.