Series-elastic actuation prototype for rough terrain hopping

In this paper, we describe development and modeling of a prototype hopping robot. The objective of our work is to create a test platform to verify control theory for fast, legged locomotion with limited sensor knowledge of upcoming, rough-terrain characteristics. Resulting running gaits aim to maximize the magnitude of unanticipated height variation, Δh, for which the hopper prototype can maintain reasonable control authority of the next apex state (and subsequent foothold). Toward quantifying controllability, we identify and quantify the range of states that our actuation strategy allows us to reach at the next apex height. We present the simplified model we use to calculate these reachable regions and provide corresponding system identification results, to justify the modeling assumptions used. This paper provides the following contributions to the field of practical robotics. First, we quantify the effectiveness of a real-world series-elastic actuation (SEA) strategy in allowing for simultaneous foothold planning and mitigation of perturbations. Second, we report on the deployment of a proof-of-concept, legged robot prototype being developed to test theoretical stance-phase control strategies.