Alex Chang
Robotics Ph.D. Student
School of Electrical and Computer Engineering
Georgia Institute of Technology

Date: April 7, 2020
Time: 10:00 AM (EST)
Location: https://bluejeans.com/5583357729 

Committee:
Dr. Patricio Vela (adviser)
Dr. Eric Feron
Dr. David Hu
Dr. Erik Verriest
Dr. Ye Zhao

Abstract:
Snake-like robots are a particular class of mobile robots uniquely characterized by their elongated, limbless bodies; this morphology and the locomotive strategies they employ permit them to traverse challenging scenarios that other robotic platforms may fail to. Gait modeling for locomotive planning and control, however, is challenged by the inherent hyper-redundancy and non-holonomicity associated with these robotic mechanisms. In this proposal a shape-centric, continuous body approach to dynamical modeling and control of rigid-link, articulated snake-like robots is explored. Gait shapes are modeled as planar continuous body curves, defined with respect to an additional average body curve and rigidly attached body frame, each residing in the plane of locomotion. Ground contact patterns encode vertical lift of robot body segments off the ground plane, in this planar model. Utilizing this framework, a traveling wave rectilinear gait is modeled dynamically, resulting in an integral closed-form, second order system. In lieu of this representation, ensuing averaged steady behavior motion characteristics of the gait may be captured as a function over the gait's parameter space. This control-to-action map effectively encodes an approximating kinematic, unicycle-like motion model of the gait; applications to trajectory planning and tracking control are then demonstrated in obstacle-strewn, planar environments. Extension of this modeling and control approach to more complex gaits, including lateral undulation and sidewinding, are proposed.