Abstract The reflex arc is fundamental to the physiology of posture and robust locomotion. Mimicking biological reflexes in engineered systems has, therefore, high potential to improve robustness and learning performance, especially when controlling humanoid robots and wearable assistive devices driven by muscle-like soft actuators. We developed an anthropomorphic, single-legged biorobot driven by pneumatic artificial muscles (PAMs) that replicates the fastest reflex in the human leg: the monosynaptic reflex loop. We achieve precise replication through a bioinspired artificial muscle spindle using the same mechanism to trigger sensory signals as humans. Defined impacts on the patellar tendon stretch the PAMs, causing the neural controller to modulate stimulation using this feedback. Through systematic analysis of the dynamics and neuronal feedback underlying biological reflex responses in 14 healthy subjects, we calibrated the free parameters of the reflex controller of our robotic leg such that its behavior is indistinguishable from humans. Our findings demonstrate the successful reproduction of the monosynaptic reflex. Moreover, the mechanism enables the implementation of bioinspired reflex characteristics in muscle-like, soft actuation devices. This allows the emulation of low-level sensorimotor control mechanisms characteristic of biological muscles, enabling engineered muscle-driven systems to exploit the inherent robustness and adaptive motor capabilities observed in biological locomotion.
Nadler et al. (Mon,) studied this question.