Managing Limited Sensing Resources for Mobile Robots Obstacle Avoidance
Juan Carlos, Cosme Rafael, Diego Álvarez, Marco Antônio · Pro Literatur Verlag, Germany eBooks · 2005
As the sensorial capacity of the robots increases, for example with the availability of MEMS sensors, it will be a requisite for good real-time performance to wisely select and manage them, in order to fulfill the strong time requirements of machines moving in human environments. This work represents a first approach towards solving this kind of problems. An analysis is presented which permits the selection of the sensor requirements for collision avoidance tasks of mobile robots. The design is compatible with motion in real time, as only the indispensable environment zones are explored, avoiding unnecessary velocity reductions. Perception system restrictions can be considered in the strategy. The analysis is deterministic, and the effect of sensor uncertainties is not discussed. The sensor model presented is general enough to be implemented with various different sensing devices. In practice, some restrictions in the election of the parameters will depend on the specific sensor characteristics, e.g., with a sonar ring, transducers are arranged around the robot with fixed intervals k, and only those pan angles are permitted. Using computer vision for range measurement gives us more flexibility, but restrictions will exist on the other two parameters. Notwithstanding, the ideas seem to be extensible to more complex models of robots or sensors, and to different motion control policies.