Real-time simulation environment for evaluating interactions between a rescue robot and rubbles

Abstract

Rescue robot simulation is an effective method for evaluating the performance of rescue robot in various rubble fields. However, simulating a rescue robot in a large rubble field such as a collapsed house needs a lot of time, since it needs to process a large number of rigid bodies and joints. This study accelerates the rescue robot simulation by using multi-core CPU and GPU on a physics engine. To consider the effect of GPU about rescue robot simulation, this study measures the processing time of two situations, including only rescue robot or a rescue robot and rubbles, on two computers equipped with different GPU. Result of the experiments show that the use of GPU, especially high-speed GPU, is effective for a situation which includes a rescue robot and rubbles. Based on the results, this study simulates a rescue robot in a collapsed house in real-time using GPU.