Characterizing Tether Friction on Natural Objects for Robotic Teams*

In field robotics, it is commonly assumed that tether-environment contact is an adversarial circumstance that should be avoided. While tether contact can increase the complexity of planning and control, harnessing the capstan effect provides an opportunity to increase the strength of robotic systems and apply larger loads. In this work, we discuss our published research intended to bridge this gap between physical complexity and function. The capstone of this work is the demonstration of a small (less than 0.5kg) legged robot on sand, detritus, and smooth low-friction surfaces supporting the tugging of more massive loads (4.8 kg).

One physical contribution in this work is the critical evaluation of the simplistic capstan equation, and its applicability to realistic unstructured settings. For example, we measure the tension amplification of wraps around redwood trees and rocks that do not have regular shapes or textures. We find that the capstan equation provides accurate estimates despite this irregularity, and can provide a realistic lower bound of predicted tension amplification. More formally, we test the ability to predict tension amplification for multiple wrapped objects in series, as opposed to all tether wrapping on a single object. We find that the capstan equation extends remarkably well to multi-capstan and other more complex scenarios, and as a result provides useful estimates to inform real-world robotic implementation.