Evaluation and Enhancement of Common Simulation Methods for Robotic Range Sensors

Distance sensors are an important class of external sensors used in many autonomous robots. Thus it is of importance to provide proper simulation for these sensors to enable software-in-the-loop testing of a robot's control software. Two different methods for distance calculation are commonly applied for the simulation of such sensors, namely reading back the depth buffer from 3D renderings and the calculation of ray-object-intersections. Many simulators impose restrictions on either method, none of the widely used robot simulators reviewed in this paper currently considers material dependent simulation of the distances measured. In this paper extended versions of both methods are presented which provide additional information on the object perceived by distance sensors. Several methods for incorporating distance- and object-information into a complete distance-sensor simulation-module are discussed. Implementations of either method are compared for their performance depending on the sensor resolution on different computer systems. These measurements show, that the break even of the methods strongly depends on the hardware, thus stressing the importance of providing either method in a robot simulation in a transparent way in order to obtain optimal performance of the simulation.