Real-Time Obstacle Distance Measurement Independent of Illumination with Monocular Camera and Cross Laser

• Published in: CACS International Automatic Control Conference (Online) pp. 1 - 6
• Main Authors: Kodama, Tomu, Mitsuhashi, Hayato, Itami, Taku
• Format: Conference Proceeding
• Language: English
• Published: IEEE 31.10.2024
• Subjects: Accuracy; Cameras; Disasters; Distance measurement; Lighting; Measurement by laser beam; Real-time systems; Reliability; Sensors; Shape
• ISSN: 2473-7259
• DOI: 10.1109/CACS63404.2024.10773154

Natural disasters have sudden and significant impacts on lives and property. To mitigate these effects, the deployment of autonomous disaster robots is increasingly seen as a critical component of efficient disaster response. A key technology in this context is obstacle detection, essential for autonomous navigation. This paper proposes a novel method utilizing a monocular camera and laser light to detect obstacles by capturing laser light features. This approach offers high reliability and cost-effectiveness, making it well-suited for disaster robot operations. However, the accuracy of distance measurement can be affected by illuminance, especially in bright environments, where the previous methods were primarily designed for use in low-light conditions. Therefore, in this study, a distance measurement method that is independent of illuminance and can accurately measure distances even in bright environments was devised. Specifically, when extracting the laser light, binarization processing was performed using YUV, a color space report that can consider luminance and hue separately, and similar parts were searched for and extracted from the laser light shape by template matching. The system succeeded in detecting obstacles with a maximum error difference of 1.3 cm.