低基線長型ステレオビジョンを用いた車両前方環境認識

概要

論文の詳細を見る
It is important to recognize surrounding environment from on-board vehicle sensors for driving support system. A stereovision system is among the most practical approaches to this problem, and we have reported on an obstacle detection method using short-baseline stereovision system, which can be installed behind the rear-view mirror of the vehicle. However, there were problems that it produced some false positives, and deteriorated measurement accuracy. In this report, we propose an obstacle detection method using Occupancy Grid Maps to solve these problems. Moreover, a detection method of moving objects is also proposed. From some experiments, it was confirmed that problems arising from shortening baseline length of stereovision were improved. Moreover, moving object can be detected from far position even though the short-baseline stereovision system is used.