Abstract
While the animals' Fin-to-Limb evolution has been well-researched in biology,such morphological transformation remains under-adopted in the modern design ofadvanced robotic limbs. This paper investigates a novel class ofoverconstrained locomotion from a design and learning perspective inspired byevolutionary morphology, aiming to integrate the concept of `intelligent designunder constraints' - hereafter referred to as constraint-driven designintelligence - in developing modern robotic limbs with superior energyefficiency. We propose a 3D-printable design of robotic limbs parametricallyreconfigurable as a classical planar 4-bar linkage, an overconstrained Bennettlinkage, and a spherical 4-bar linkage. These limbs adopt a co-axial actuation,identical to the modern legged robot platforms, with the added capability ofupgrading into a wheel-legged system. Then, we implemented a large-scale,multi-terrain deep reinforcement learning framework to train thesereconfigurable limbs for a comparative analysis of overconstrained locomotionin energy efficiency. Results show that the overconstrained limbs exhibit moreefficient locomotion than planar limbs during forward and sideways walking overdifferent terrains, including floors, slopes, and stairs, with or withoutrandom noises, by saving at least 22% mechanical energy in completing thetraverse task, with the spherical limbs being the least efficient. It alsoachieves the highest average speed of 0.85 meters per second on flat terrain,which is 20% faster than the planar limbs. This study paves the path for anexciting direction for future research in overconstrained robotics leveragingevolutionary morphology and reconfigurable mechanism intelligence when combinedwith state-of-the-art methods in deep reinforcement learning.