Structured Light Vision-Based Automatic Gripping of Pipes in Biped Climbing Robots

The gripping operation is paramount for biped climbing robots (BiCRs), as an improper grip may result in high-rise falls. Automating this operation contributes to expediting grip establishment and enhancing grip quality. In this study, we present a comprehensive and versatile scheme for enabling automatic grip operations in BiCRs. Considering grip constraints and offline grip planning results, we identify the requisite parameters of featureless cylindrical pipes for calculating a proper grip. To enable online estimation of the corresponding parameters, we propose a sensing system that integrates two line lasers and a camera, along with algorithms featuring one-shot query and iterative optimization capabilities with varying levels of accuracy and consumption time. Experimental results affirm the commendable performance of our sensing system concerning accuracy, efficiency, and robustness to working range and illumination variance. Additionally, a multistep climb experiment of Climbot on a scaffold validates the efficacy of our proposed automatic gripping scheme. This scheme proves advantageous in practical applications that require BiCRs to utilize offline planning results and online perception information to perform long-distance movements.