An Energy Efficient Rimless Wheel

Sunday, February 17, 2013
Auditorium/Exhibit Hall C (Hynes Convention Center)
Mario W. Gomes , Rochester Institute of Technology, Rochester, NY
Geoff Bastian , Rochester Institute of Technology, Rochester, NY
Background:
As mobile robots become more integrated into our daily lives, the amount of energy that they use will be a critical impediment to their usefulness. Bipedal robots are notoriously inefficient walkers when compared to humans . Cost of transport (COT) is an often used measure of locomotion energy efficiency and is defined as [COT = energy used/(weight x distance)]. One of the most efficient walking robots is the Cornell Efficient Biped which has a cost of transport of 0.2, similar to human walking. It is known that energy losses due to collisions are the primary mechanical energy loss mechanism for dynamic walking robots.
The primary objective of our work is to understand the dynamics of a non-wheeled, mobile robot which eliminates the primary mode of energy loss for the most efficient walking robots, namely, energy lost through collisions between the foot and the ground. This efficiency is achieved by creating a system that is capable of motions where the foot contacts the ground with zero relative velocity. The 2D system we examined consists of a rimless wheel coupled to an inertia wheel via a torsional spring.

Methods:
A 2D, 2 DOF model of a modified rimless wheel system was constructed, and the coupled non-linear equations of motion were derived for that system. Solutions to these equations were then approximated using standard explicit numerical integration routines. The initial condition and parameter space was explored to find a set of buildable parameters. This system was then built and tested.

Results:
We have shown through numerical simulations that it is possible to carefully design a modified rimless wheel, to traverse a flat floor with the same energy-loss as a rolling wheel. We have built a physical device which can demonstrate the fundamental mechanism of collision-less motions for energy-efficient locomotion of walking robots. Video analysis of experimental testing showed that the unactuated, uncontrolled, prototype has asymptotically stable periodic "walking" motions down an 8 degree ramp angle. An 8 degree ramp angle corresponds to a COT of approximately 0.14. We have also found, in simulation, workable control algorithms which would stabilize the half-stable zero-cost periodic motion across level ground.

Conclusions:
The technique of collisionless walking, and the physical testbed which demonstrates it, could naturally lead to the development of walking robots which use much less energy to move about. We predict that COTs less than 0.05 may be possible for robotic walkers which use this technique.