Versatile sensing architecture for autonomous mobile robots

This paper presents a software architecture that can be used in different robots, each characterized for its own set of sensors. The versatility of the architecture depends on the capability of the sensing subsystem for working with different types of sensors. With the purpose of implementing a system easy to reconfigure the architecture has been designed and implemented using a real time object-oriented paradigm called ROOM. Not only does ROOM let users implement a versatility sensing architecture for different sets of sensors, but also many other scenarios. Different sensor fusion algorithms can be tested inside the software architecture with the same sensors, the same system can be used in different robots (real or simulated), the control architecture is independent on the target and operating system,.... The properties of the architecture are illustrated with a detailed explanation of different sensing architectures implemented in the control system. The last sensing subsystem shows a unified fusion system based on Bayesian Networks for estimating the robot location and building a map.