Nonlinear, muscle-like actuation reduces energy consumption— does it also simplify control?

Organisms convert chemical energy to mechanical work, allowing them to perform a wide range of impressive mechanical tasks. The most common organ that performs this kind of conversion is muscle. Seminal work by Hill demonstrated that muscle is inherently nonlinear, with a hyperbolic force-velocity relation. This is in stark contrast to electromagnetic motors— common actuators in household appliances and robots— which have linear properties. What are the mechanical advantages of nonlinear actuation? Answering this question is difficult, because force-velocity relations of living muscle cannot be systematically varied. Here, we seek to answer this question with "HillBot", a robot that lifts a weight against gravity while mimicking muscle's nonlinear force-velocity relation using feedback control. We systematically vary a parameter, α, that controls nonlinearity. Even after accounting for the lower power characteristic of high-α actuators, we find that increasing α decreases energy consumption. We predict that nonlinear actuation determines a tradeoff between economy and performance. We furthermore hypothesize that nonlinearity improves robustness to perturbation and simplifies the computational burden of neuromuscular control.