A control architecture of autonomous robots and its realization using multiple DSP devices and RTOS based platform

This paper presents a control architecture for robots in dynamic and unstructured environments. It is based on the combination of several parallel behavior modules: reactive behaviors and directive behaviors with path planning and behavior state feedback, which makes behavioral closed loop. A finite state automata model is used to describe the activity sequence the robot must follow. Detailed behavior planning including path planning is executed by stages to prevent the computation waste and the sensing function is activated whenever possible. At last, a distributed computational mechanism of multiple DSP devices is developed and functional modules are implemented in the way of multitask on a RTOS platform. The system has an open modular and scalable architecture, and a high performance with a low cost.