Mass-Constrained Robotic Climbing on Irregular Terrain

Climbing robots can operate in steep and unstructured environments that are inaccessible to other ground robots, with applications ranging from the inspection of artificial structures on Earth to the exploration of natural terrain features throughout the solar system. Climbing robots for planetary exploration face many challenges to deployment, including mass restrictions, irregular surface features, and communication delays. We present a hierarchical approach that overcomes these obstacles via underactuated design to comply with surface features, internal force control to maximize adhesion, and hybrid path planning to enable autonomy on complex 3D terrain. We first apply compliant mechanism design at the scale of individual microspines, which rely on flexible suspensions to load-share and conform to surface features. We present a new suspension design that can be 3D printed from a single material to enable faster and easier fabrication with no reduction in mechanical performance. We then apply underactuation at the gripper scale, presenting a fully passive microspine gripper and wrist that can nevertheless solve the paired problems of conforming to uneven terrain and maintaining adhesion over a wide range of loading angles. These grippers are mounted on the lightweight rock-climbing robot LORIS, which has demonstrated the successful ascent of several different climbing surfaces. Next, we move to the full robot scale with the design of Sally, a magnetic-wheeled inspection robot for steel structures. Sally uses a passive suspension to maintain end effector contact despite surface irregularities and repurposes the steering and sensor deployment actuators to assist in difficult corner transitions. At the control level, we generalize the bioinspired directed inward gripping technique as an internal force optimization problem that we apply to both LORIS and Sally. We also present a unified shape, contact, force, and motion control scheme that was used by the snake-like robot EELS to successfully descend an ice shaft. At the planning level, we present a motion planning approach for hybrid locomotion that applies the sampling-based AIT* algorithm to rapidly search for paths along arbitrarily shaped terrain manifolds modeled by point clouds. This global planner relies on a local planning algorithm to select high-quality footholds. Lastly, we examine planner solutions both on hardware and in simulation. We end by discussing possible extensions of this work to conventional quadrupeds and the pursuit of dynamic climbing capabilities.