A Novel Model for Link Dynamics in Planar Snake Robots Using Internal Constraint Force Sensing

Abstract

We consider the problem of simplifying the typically complex task of deriving dynamical mechanical models of planar snake robots. More precisely we propose a modeling strategy that assumes the possibility of measuring the constraint forces acting between adjacent links in a snake robot, something that is now technologically possible thanks to currently available compact commercial sensors. We show that this information can be used to decouple the dynamics of each link in the snake robot to build a novel dynamic model that is simpler than the typical models in the available literature, but still powerful for predicting the movements of the robots. The proposed model may help to significantly reduce the computational complexity associated with model-based control and estimation schemes compared to other established models. Besides this, we show that the proposed model exhibits multiple properties that ease performing control, identification and state estimation tasks in general. More specifically, we show that parts of the dynamics of the model can be considered to be linear with a known nonlinear exogenous disturbance which can be eliminated using feed-forward control. Finally, we show that linear Kalman Filters remain the best linear unbiased estimators for part of the state even when exogenous disturbances enter non-linearly in the system.

A Novel Model for Link Dynamics in Planar Snake Robots Using Internal Constraint Force Sensing