Hierarchical Reactive Task Planning with Temporal Logic and Visual Servoing for Bolt-Tightening Robots in Transmission Towers

Abstract

With the rapid deployment of unmanned bolt-tightening robots on transmission towers, traditional control algorithms face challenges in balancing long-term task logic and real-time adaptability, especially under unstructured environments such as missing bolts and unexpected obstacles. This paper proposes HTP-TV, a framework for hierarchical task planning with temporal logic and visual servoing, which integrates temporal logic-based planning with a vision-based reactive mechanism. HTP-TV decouples semantic goals such as bolt-tightening sequences from geometric path planning, enabling offline pre-planning via LTL-RRT* to generate constraint-compliant trajectories. In the online phase, real-time camera data dynamically updates the environmental model, which triggers adjustments in the incremental Büchi automaton to address missing bolts or obstacles. A semantic ID system encodes the bolt topology, supporting re-planning with axial constraints, while visual servoing techniques correct execution deviations. Through comparisons with the two methods (Offline LTL-RRT* and Online RRT*-based), the simulation results in CoppeliaSim demonstrate the efficiency, high safety compliance, and superiority of HTP-TV.