Investigating the Performance of Corridor and Door Detection Algorithms in Different Environments

Abstract

The capability of identifying physical structures in an unknown environment is important for autonomous mobile robot navigation and scene understanding. A methodology for detecting corridor and door structures in an indoor environment is proposed, and the performances of the corridor detection algorithm and door detection algorithm applied in different environments are evaluated. In the proposed algorithms, we utilize a feedback mechanism based hypothesis generation and verification (HGV) method to detect corridor and door structures using low level line features in video images. The proposed method consists of low, intermediate, and high level processing stages which correspond to the extraction of low-level features, the formation of hypotheses, and the verification of hypotheses using a feedback mechanism, respectively. The system has been tested on a large number of real corridor images captured by a moving robot in a corridor. The experimental results validated the effectiveness and robustness of the proposed methods with respect to different viewpoints, different robot moving speed, under different illumination conditions and reflection variations.