A Distributed Network of Microcontrollers for Mobile Robots

Mobile robots are often equipped with one powerful μC or PC managing a large system with sensors, actuators and tasks. The contemporary approach is to divide functionality into several smaller sub-systems. The main advantages of such a decentralized and dynamic system are better maintainability, higher reliability, improved fault tolerance, and a more flexible distribution of tasks. Therefore, this paper presents a decentralized self-organizing and networking system for μC connected via CAN. Algorithms for managing joining and leaving nodes at unpredictable times, for time synchronization, for fault management as well as for automatic reorganization are integrated in the system and will be introduced in this paper. The system was implemented on a network of μCs to demonstrate the performance of the system. Finally, a sensor data fusion to determine the position of a mobile robot was implemented based on a Kalman Filter and using sensors connected to several μCs.