A Worm-like Robot for Ferromagnetic Surface Inspection

Industrial inspection is a crucial task for ensuring the long-term safety and maintenance of infrastructures with ferromagnetic surfaces, such as power plants, wind turbines, chemical facilities, and storage tanks. These environments are often hazardous or inaccessible for human operators, which has motivated the development of climbing robots capable of performing inspection tasks reliably and repeatedly. However, many existing robotic systems suffer from limited maneuverability, reduced stability during surface transitions, and high energy consumption.In our publication in IEEE Robotics and Automation Magazine, we introduced a novel climbing robotic system inspired by the body morphology and locomotion of caterpillars. The design consists of three modules connected via a spring-joint mechanism and equipped with a passive magnetic adhesion system. The system’s maneuverability, stability, energy efficiency, and robustness were studied through both analytical modeling and extensive experimental validation. Experimental results demonstrated the robot’s ability to perform complex maneuvers, including internal/external corner transitions, thin-surface transitions, pipe circling, and surface turning, under real-world conditions such as rusty, coated, and uneven surfaces. Furthermore, we showed that the proposed bio-inspired worm-like locomotion achieves an average 10.2\% increase in forward velocity and 6.7\% reduction in the cost of transportation and energy consumption compared to simple locomotion strategies.We use this Figshare repository to openly share our experimental videos and supplementary materials. Each video documents a specific locomotion mode, surface transition, maneuverability test, or targeted validation of robustness and energy efficiency.Based on our modeling, we categorize surface transitions according to their orientation relative to gravity. In our experiments, gravity direction is explicitly considered, and for each transition category, we evaluate three representative motion cases. For example, for inside-corner structures, three cases are defined based on the robot’s direction of motion relative to gravity; the same categorization is applied to other structures, such as outside corners and pipes. When gravity direction introduces asymmetric challenges (e.g., floor-to-wall or ceiling-to-wall transitions), both upward and downward motions are tested. For wall-to-wall transitions, where gravity effects are symmetric, a single representative case is sufficient.Video 1 – Experiment 1 (Inside Corner Transitions) presents various inside-corner transition configurations. Additional tests in this scenario include transitions involving structural features such as ramps, obstacles, and non-ferromagnetic indentations beneath the surface, reflecting realistic industrial conditions. Since the robot successfully handled these features in other experimental scenarios, they were not repeated in subsequent experiments.Video 2 – Experiment 2 (Outside Corner Transitions) follows a similar structure, evaluating outside-corner transitions under three gravity-relative configurations. One of the most challenging cases—transitioning from ceiling to wall—is shown, where the robot successfully completes the maneuver through module coordination and compliant spring joints.Video 3 – Experiment 3 (Turning) focuses on maneuverability in terms of turning performance. Owing to the compliance of the spring joints, the robot is capable of executing complex turning behaviors, including smooth spiral trajectories.Video 4 – Experiment 4 (Pipe Circling and Climbing) demonstrates the robot’s ability to adapt passively to curved structures. Both gravity-aligned and gravity-opposed movements are tested, as well as spiral motion around a pipe, highlighting the benefits of modular compliance and the passive adaptation of each module.Video 5 – Experiment 5 (Robustness) shows the system’s fault tolerance. The experiments show that a single module can support and recover the robot under scenarios involving complete detachment of up to three modules or partial detachment of two modules, preventing catastrophic failure.Video 6 – Experiment 6 (Energy Efficiency) presents the bio-inspired locomotion strategy designed to minimize the cost of transportation. Extensive experiments were conducted to identify the optimal locomotion frequency, and the video shows the configuration yielding the best performance.Video 7 – Experiment 7 (Surface Roughness and Real-World Conditions) evaluates the robot on surfaces with varying magnetic adhesion due to rust, dust, coatings, and surface irregularities, demonstrating reliable performance under realistic industrial conditions.Video 8 – Experiment 8 (Thin Wall Transitions) presents transitions across surfaces with minimal width, a particularly challenging scenario. In this configuration, the robot effectively bypasses the intermediate thin surface and transitions directly to the adjacent surface; as a result, forward and backward motions exhibit similar behavior and are therefore not distinguished.Finally, the Experimental Verification of the Mathematical Analysis section includes internal force measurements and validation experiments supporting the analytical models presented in the paper.