Meeting Abstract
Animals rely on local sensory feedback to maintain a variety of complex leg/foot trajectories when navigating natural terrains. The choice of trajectory is determined by a combination of factors including to body morphology, actuation capabilities, performance requirements, and environment. In contrast, a majority of bioinspired legged robots today are typically restricted to utilizing one (or few) pre-programmed leg motions during running due to the complexity of leg design, difficulty in actuation, and a lack of reliable sensing resulting. To address this issue, we present the latest generation of the Harvard Ambulatory MicroRobot (HAMR) – an insect scale (45 mm long, 1.43 g) quadrapedal robot that retains mechanical complexity despite its small size. HAMR is capable of high-speed locomotion on level ground, can climb vertical and inverted walls, and even swim on the water’s surface. In order to test the hypothesis about leg trajectories affecting locomotion performance, we have developed and integrated a novel motion encoder that provides a reliable estimate of the robot’s joint position and velocities. We then utilize this proprioceptive feedback to control heuristically designed leg trajectories and demonstrate that we can recover locomotion performance (stride length) in highly dynamic frequency regimes (10-30 Hz). Additionally, at higher frequencies (40-50 Hz), we observe that the shape of leg trajectories is less important if the energy exchange between the robot and terrain is appropriately modulated. With precise control over arbitrary leg trajectories, we can now begin to test hypotheses about the choice of leg trajectories in biological systems at scale.
