A simple network of proprioceptive reflexes can produce a variety of bipedal gaits
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Proprioception is crucial for movement, yet the role of proprioceptive reflexes in legged locomotion is still poorly understood. While previous simulation studies have shown great potential for reflex-based control strategies, these controllers are typically catered to specific gaits, using hand-crafted feedback pathways that are linked to specific gait phases. In this work, we explore the control capabilities of a simple reflex controller that consists of only monosynaptic and antagonistic length and force feedback pathways with constant gains. Despite its simplicity, we found our control framework capable of producing a wide variety of natural gaits, including walking and hopping, forwards and backwards, and running in different variations and at different velocities – without requiring any rhythmic inputs or high-level state machines modulating the feedback gains. Our work highlights the important role and flexibility of proprioceptive reflexes and suggests a necessary re-evaluation of their role in locomotion. Due to its simplicity and flexibility, our control framework provides a solid basis for the development of higher level neuromuscular control systems.
Author summary
Human locomotion is characterized by an impressive versatility and humans transition naturally between different gaits. However, the neural mechanisms underlying human motor control are not fully understood and remain controversial. The two widely accepted basic components of the neural circuitry involved in locomotion are central pattern generators (CPG) and spinal reflexes, leading to a long-standing debate on their respective roles. Neuromusculoskeletal models can be employed to study the role of these neural control structures in simulation. Until now, these models mostly focus on walking and require an additional component like a state-machine or CPG besides reflexes to generate rhythmic activation of muscles. In this work, we propose a controller that can generate different rhythmic gaits like walking and hopping, forwards and backwards, as well as running, based solely on simple reflexes known to exist in the mammalian spinal cord. We challenge the current view of the role of rhythm-generating components and show the remarkable potential of reflexes for locomotion control. To our knowledge, this is the first work showing that versatile locomotion can be achieved solely based on reflexes without the need of a CPG. Our results contribute to the understanding of human motor control and are relevant for the control of robots, exoskeletons, prostheses and ortheses.
