PRE-PRINT A robot that walked 65 km on a single charge: energy-effective and reliable locomotion using trajectory optimization and stabilization from reflexes∗

No legged walking robot yet approaches the high reliability and the low power usage of a walking person, even on flat ground. Here we describe a robot which makes small progress towards that goal. Ranger is a knee-less 4-legged ‘bipedal’ robot which is energetically and computationally autonomous, except for radio controled steering. It has a total cost of transport [TCOT ≡ P/mgv] of 0.19, less than a human’s TCOT ≈ 0.3. Ranger walked 65.2 km in 186,076 steps in about 31 hours without being touched by a human (using a not yet fully optimized controller with a TCOT of 0.28). The high reliability and low energy use are achieved by: 1) development of an accurate bench-test-based simulation; 2) numerical trajectory optimization using that simulation; 3) a simplified trajectory optimization to ease implementation; and 4) offline design of a simple event-driven discrete feed-forward stabilizing controller. At TCOT = 0.19, the robot’s total power of 11.5 W is used by sensors, processors and communications (45%), motor dissipation (≈34%) and positive mechanical work (≈21%). Ranger’s reliability and low energy use suggests that simplified implementation of offline trajectory optimization, stabilized by a low-bandwidth reflex-based controller, might lead to energy-effective reliable walking of more complex robots. ∗Other candidate titles: The first robot to walk a marathon A robot that can walk far using little energy Energy-effective, reliable 2D walking via trajectory optimization and reflex based control An energy-effective and reliable walking robot via trajectory optimization and reflex based control A robot that can walk far using little energy using trajectory optimization and event-based stabilization Funding from NSF grant 52836 to AR.