Qinrui Tang

Autonomous mobile robots (AMRs) face challenges in navigating complex environments efficiently. To manoeuvre through both narrow and wide spaces, AMRs require two essential design features: a compact form for tight areas as well as a large configuration with omnidirectional movement for wide spaces. This study utilizes inhouse designed Expand and Collapse Variable Width Robot (ECVWR) to demonstrate effective area coverage. These ECVWR can contract to navigate constrained spaces and expand to optimize coverage in open areas. Existing methods for achieving complete area coverage do not account for the reconfiguration or the change in footprint. To address this issue, we propose the Depth-First Search (DFS) for Complete Coverage Path Planning Strategy (CCPPS) for ECVWRs. This method allows adjustments to the generation of waypoints in CCPPS, minimizing the path length. The simulation study shows that our proposed CCPPS outperforms contemporary state-of-the-art CCPPS, namely, GBNN, exhibiting superior expanded area coverage, reduced travel distance, and enhanced computational efficiency. Moreover real-world experiment to further benchmark the efficacy of our proposed algorithm. The proposed CCPPS is generic and can be extended to other variable footprint robots.

Complete coverage path planning (CCPP) is essential for autonomous cleaning robots, particularly in complex and variable environments where traditional, fixed-footprint designs may fall short. This paper presents adaptive Glasius bio-inspired neural network (aGBNN) approach to CCPP, specifically tailored for an omnidirectional self-reconfigurable cleaning robot (OSCR). The aGBNN method dynamically generates a complete coverage path by leveraging the ability to change sweeping footprint of the robot (SFR) assisted by reconfiguring brushes design. The sweeping is carried out both longitudinally and laterally, thereby complementing the omnidirectional locomotion with cleaning. Unlike conventional CCPP algorithms that assume a fixed robot footprint, the proposed aGBNN adapts in real-time to spatial and moving obstacles, significantly enhancing coverage efficiency. Experimental and simulation results demonstrate the advantage of the aGBNN approach, in terms of path length, and total time to complete area coverage compared to state-of-the-art methods.