Enabling New Mobility Capabilities for Robots Under 100g Using Scaling Laws and Comparative Morphology

Carnegie Mellon University, Pittsburgh PA

Investigators

Abstract

This Foundational Research in Robotics (FRR) project will develop the world’s smallest functional bipedal robot, multi-legged robot, and wheeled robot. It will be a hundred times smaller by mass than the current state of the art, developed by studying the scale-dependent differences in actuation, mechanics, and contact forces. Specifically, this new family of robot designs will include the first bipedal robots capable of walking, turning, and stepping onto obstacles at sizes below 1 gram. It will also be the first fully controllable multilegged robots at the 10mg scale, and the smallest externally actuated wheeled robot at the 1mg scale. Small mobile robots are useful for accessing tight, confined spaces, including for industrial inspection, micromanufacturing, and exploring rubble. However, existing small robots have not been able to locomote nearly as well as larger robots. The sensing, actuation, computation, and manufacturing techniques available are more limited than at larger scales. Small systems have relatively higher friction, damping, and other losses that they must overcome. These differences and limitations will affect different robot morphologies in different ways, and so it is not clear that the same locomotion strategies that work well for large robots will scale down to smaller ones. A learning module called the "Robot Zoo" will take advantage of the wide variety of robot shapes, sizes, and locomotion schemes used in this study. This module will be integrated into existing K-12 outreach programs at Carnegie Mellon University. In order to achieve new robot designs and capabilities, the project will develop a set of scaling laws that describe how components and systems behave at different size scales, with a focus on actuation, joint losses, and contact properties. These scaling laws will be used to determine how designs will have to change at smaller sizes, e.g. how large the actuators must be to overcome the additional losses, as well as what design strategies will be most efficient, e.g. comparing different joint structures. These results will be used to make new, smaller designs for bipedal, multilegged, and wheeled robots with improved capabilities in terms of speed, efficiency, and maneuverability. The project will also develop metrics to quantify the relative performance of different robot designs across a range of properties. These metrics will be used for comparative morphology to better understand the relative tradeoffs between different designs and, ultimately, answer the central question of this project: What is the best way to move at small scales? This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.