影のある環境下での自律ロボットの道路認識システムの性能向上
Enhancement performance of road recognition system of autonomous robots in shadow scenario
影が道路認識に与える影響を調査し、影を検出・除去するアルゴリズムを導入して自律走行車の道路領域認識精度を改善した研究。
著者: Olusanya Y. Agunbiade, Tranos Zuva, Awosejo O. Johnson, Keneilwe Zuva
分類: cs.RO, cs.CV
原文アブストラクト
Road region recognition is a main feature that is gaining increasing attention from intellectuals because it helps autonomous vehicle to achieve a successful navigation without accident. However, different techniques based on camera sensor have been used by various researchers and outstanding results have been achieved. Despite their success, environmental noise like shadow leads to inaccurate recognition of road region which eventually leads to accident for autonomous vehicle. In this research, we conducted an investigation on shadow and its effects, optimized the road region recognition system of autonomous vehicle by introducing an algorithm capable of detecting and eliminating the effects of shadow. The experimental performance of our system was tested and compared using the following schemes: Total Positive Rate (TPR), False Negative Rate (FNR), Total Negative Rate (TNR), Error Rate (ERR) and False Positive Rate (FPR). The performance result of the system improved on road recognition in shadow scenario and this advancement has added tremendously to successful navigation approaches for autonomous vehicle.
