Autonomous mobile robot simulator-a programming tool for sensor-based behavior

An autonomous mobile robot must achieve its goal in very complex environments with uncertainties of sensors and actuators. Due to such uncertainties, the control algorithm of robot behavior must have the ability to cope with various possible environmental situations and robot status. To develop such a control algorithm of robot behavior, the algorithm must be tested under numerous conditions of the robot's environment. Such a process requires a large number of experiments using real robots and because of high experimental cost and environmental complexity, a realistic simulator should be developed for verification of behavior algorithms. In this paper, we demonstrate the necessity and usefulness of an autonomous mobile robot simulator as a programming tool which simulates all robot functions and environments including dynamic motion of a robot, control software of robot's subsystems, sensor characteristics and behavior level software. And we point why such a simulator can act as the center of a programming environment for developing robot behavior algorithms. Accordingly, we describe Autonomous Mobile RObot Simulator (AMROS) which is developed as a programming tool for sensor based behavior. AMROS consists of simulation of vehicle controller process, simulation of vehicle motion based on dynamics model, simulation of ultrasonic range sensor, simulation of ROBOL/0 behavior program execution and simulation of indoor environment. To realize AMROS, synchronization method among all parts of the simulation is considered. Synchronization mechanism that a behavior description language ROBOL/0 has, is utilized for this synchronization. Due to the fact that sensory information is the only way to know environmental conditions, a realistic simulation of sensor interaction with robot's environment is necessary. Based on this concept, an ultrasonic range sensor simulator, which simulates propagation process of ultrasonic wave, is developed and described in this paper. AMROS targets the real mobile robot "Yamabico" operating in an indoor environment. The efficiency of the results obtained through simulation are presented by comparing to the results obtained by real experiment. Lastly, we present our experience of implementing behaviors of the mobile robot with some examples that show the high performance of the developed simulator.