Title: Neuromechanics And Augmentation Of Muscle-Tendon Actuators In Unsteady Cyclic Tasks 

Date: Friday, Dec 2nd 2022 

Time: 11AM-12 pm ET (public), 12PM-1:30PM (private)

Location: 3515 Conference Room MRDC (Zoom link: https://gatech.zoom.us/j/95570225623?pwd=LysweGVrcUZCWkRnSGtScStjNS91Zz09)

Laksh Kumar Punith 

PhD Candidate in Robotics 

George W Woodruff School of Mechanical Engineering 

Georgia Institute of Technology  

Twitter : @LakshKumarP  

LinkedIn : https://www.linkedin.com/in/lakshkumarp/ 

Committee

Dr. Gregory S Sawicki (advisor), Associate Professor, Woodruff School of Mechanical Engineering, Georgia Tech 

Dr. Lena H Ting, John and Jan Portman Professor, Wallace H. Coulter Dept. of Biomedical Engineering, Emory University and Georgia Tech 

Dr. Simon Sponberg, Dunn Family Professor, School of Physics, Georgia Tech 

Dr. Manoj Srinivasan, Associate Professor, Department of Mechanical Engineering, Ohio State University 

Dr. Jonathan W Hurst, Professor of Mechanical Engineering and Robotics, Oregon State University and Chief Technology Officer and Co-Founder of Agility Robotics 

Abstract 

Legged animals navigate complex environments with incredible stability, agility and economy despite having significant neuromechanical constraints like large delays and highly compliant actuators. They do so partly by tuning the mechanics of their actuators (i.e. muscle-tendon units) to act in a context-dependent manner. This raises several questions, three of which are discussed in this thesis. (A) to what extent can you purely rely on the mechanics of your actuators? In particular, can muscle-tendon units reject perturbations like uneven terrain without changing neural control? (B) how does stability, agility and economy vary with changing muscle-tendon properties individually and how do they tradeoff? and (C) if morphology affects movement performance in animals, can we augment human function across multiple objective functions (namely stability agility and economy) simultaneously by augmenting the morphology of muscle-tendon units with passive wearable robots. To answer these questions in a causal, controllable and generative manner, we developed a framework where a single muscle-tendon unit is interacting with a mass in gravity through a lever arm in closed loop to generate cyclic movement with variable terrain (both in simulation and in-vitro closed-loop experiments), variable morphology (in simulation) and variable nervous system control (in simulation). Through our work, we show that (A) muscle-tendon units can rapidly stabilize a hopping body when faced with a sudden change in ground height despite zero change in neural control, (B) series elastic tendons variably influence stability, agility and economy of movement such that animals need to trade off stability, agility and economy when tuning their muscle-tendon properties and (C) passive elastic exoskeletons are able to simultaneously augment stability, agility and economy despite being 'spring-like' and unable to do net work themselves by shifting the mechanics of underlying muscle-tendon units. Through our research, : (1) we gain fundamental neuromechanical understanding of how animals enable stable, agile and economic movement by tuning their actuators and (2) we generate a template for the design of a new generation of bioinspired robotic actuators to enable legged and wearable robots to navigate the world in all its richness and complexity.