Abstract
A method of recognizing the characteristics of spatial structure for an autonomous mobile robot (AMR) is proposed in this paper. The recognition scheme is based on Bayesian hypothesis reasoning, and uses ultrasonic sensor data obtained from an robot in an unknown environment. The optimal hypothesis reasoning scheme is difficult to implement due to its exponentially increasing number of hypothesis as a function of observation time. The proposed scheme uses only current spatial representative primitives (SRPs) along with their transition probabilities and approximates the optimal scheme. The SRPs are defined as spatial continuous characteristics. The scheme uses all SRPs and the Bayesian reasoning for generation of exploration direction. This is similar to that of human spatial recognition and it improves the performance of a hypothesis reasoning scheme. The performance of the proposed scheme is evaluated by testing an AMR in a building environment.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have