Abstract
In order to achieve autonomous mobile robots, the robots need to estimate current positions. Generally, mobile robots obtain sensor data from external sensors such as laser range sensors or cameras, and estimate positions based on matching with maps prepared beforehand. Global position estimation based on GPS can be also performed. However, sensor data with low reliability is often obtained depending on situations or positions. Such sensor data will make position estimation accuracy worse. In this study, dynamic sensor selection is proposed and implemented to an autonomous mobile robot. Reliability of sensor data is dynamically evaluated and suitable sensor is adopted for estimation. In this paper, simulations and experiments with an actual robot based on the proposed method are shown.
