Benjamin McInroe

The natural movement behaviors that we (and other animals) use throughout our lives demonstrate agility, robustness to disturbance, and adaptability that far surpasses those of existing synthetic systems. I'm interested in uncovering the mechanisms by which these dynamically dexterous behaviors emerge from the control and coordination of coupled neural, mechanical, and sensory systems. My current research focuses on the identification and synthesis of control modules - interpreted as low dimensional, composable representations of embodied dynamic behaviors - and their hierarchical compositions that produce the diverse motor repertoires characteristic of living systems. Towards this goal, I use an integrative approach combining biomechanics (3D motion capture and force measurement), applied mathematics (dynamical systems, control, and learning algorithms), and robotics (physics-based simulation and embodied systems).

Presently, I am a postdoc at the U Penn Department of Electrical & Systems Engineering, where I work on approaches and algorithms for learning control modules from high dimensional trajectory observation data in collaboration with Dan Koditschek and Yuliy Baryshnikov.

I completed my PhD in Biophysics at UC Berkeley with Bob Full, where I used biomechanics experiments, physics-based simulations, and robotics to reveal control modules underlying whole-body dynamic behaviors in animals. My dissertation research included identification of the spine and appendage control modules that enable small animals to self-right across terrain types without loss of performance by using multiple actuation modes, and terrain manipulation primitives that enabled the first legged burrowing robot.

Before coming to Berkeley, I completed my B.S. in Physics at Georgia Tech in 2015, where I worked with Dan Goldman on elucidating the appendage coordination and control strategies underlying the evolution of legged locomotion in terrestrial environments.