Multisensor stochastic integration and control

Abstract

Aiming to control a multisensor setup, we develop a stochastic system model as a step toward designing robotics perception systems, which state spaces show both discrete (for example presence of objects) and continuous (position, speed..) parts. We present here this model together with results in multisensor simulation. To conclude with, we comment on the emergence of symbolic techniques into the probabilistic integration and control, and we briefly describe the first implementation of the real setup control.