Design and experimental verification of a bionic eye mechanism for vision-guided roadheader motion control

Abstract

To address high accident rates, low safety, and poor remote-control reliability in tunnel excavation operations, especially in deep coal mines and other resource-extraction environments, this study proposes a bio-inspired vision-driven motion control method for roadheader system based on biological visual-perception principles and human decision-making mechanisms. The proposed method aims to enhance the adaptive precision control of the roadheader pose across all operating conditions. Firstly, a bionic eye visual system driven by dual-motor cable mechanisms was designed, by referencing human eye movement, featuring a two-degree-of-freedom platform for dynamic environmental perception. Secondly, a real-time visual pose estimation model based on the Efficient Perspective-n-Point algorithm was developed, incorporating image-preprocessing techniques to ensure rapid pose computation. Finally, a fuzzy control model that integrates the kinematic constraints of the forward motion of the roadheader and a logical control model based on cutting kinematics were established, and the effectiveness of the control models was validated through simulation. Experimental results demonstrate that the proposed autonomous cutting control method satisfies tunnel excavation requirements. Specifically, the designed bionic eye system achieves vertical rotation of ±45° and horizontal scanning of ±65°, with angular velocities exceeding 50°/s and stable stepwise angular execution with a mean relative error below 2%. In simulation experiments, trajectory control error remained within the industry standard of 50 mm, with the maximum undercutting amount of 20.694 mm and overcutting not exceeding 56.92 mm.