20-04-2010, 03:26 PM
Abstract
This paper presents a remotely controlled mobile robot able to work in an unstructured environment. The main constraint is represented by the communication time (Milan - Ploiesti), that was evaluated of two orders higher than the inner clock of the control system. This difficulty has been avoided by giving a short-term autonomy to the system. Namely, the control system was configured as a neural network on board of the robot that reacts to external stimuli. A copy of the network is in the remote control station. Here a workstation periodically identifies the system dynamics, the robot environment and the optimal control strategy. The strategy is translated in a proper selection of the free parameters of the neural network. The values of these parameters are transmitted as remote control action to the robot neural network. The paper is organized as follows. In section 2, the architecture of the robot is described, and in section 3, the kinematic equations of the robot in a fixed frame and in a mobile frame are introduced. The problem of following a wall with one or two ultrasonic pairs is modeled in section 4. The training strategy for computing reliable control signals is described in section 5 and, in section 6 the results of experiments and simulations are given.