If someone built a robot whose locomotion was modeled on that of a three-year-old kid hopped up on too much candy, it would probably act a whole lot like the University of California, Berkeley’s Salto robot.
Created by the university’s always-interesting Biomimetic Millisystems Lab, Salto (the name is short for “Saltatorial Locomotion on Terrain Obstacles”) exhibits some darn impressive leaping abilities, capable of putting lesser NBA ballplayers to shame. We’ve seen some impressive jumping robots before, but this is some next-level stuff!
“Salto is a new robot that we built to study jumping locomotion,” Dr. Duncan Haldane, a roboticist at the university, told Digital Trends. “Unlike previous robots that spend a long time winding up or just can’t jump very high, Salto can do very large jumps, one after another. Our latest improvement to Salto is a set of propellers that let Salto balance upright so it can jump around a room.”
Haldane said he was inspired to build the 98-gram Salto after paying a visit to the FEMA Urban Search and Rescue training site in nearby Menlo Park, California. “I wanted to build a small robot that could move very quickly through complex environments like collapsed buildings,” he continued. “The more general application is really high mobility for small robotic systems. With our new jumping strategy, you can take a robot that weighs less than a stick of butter and have it move around in human-scale environments.”
The Salto robot is controlled via two small thrusters that thrust either in the same or different directions, depending on whether it wants to yaw or roll — referring to the axis around which it rotates. It stabilizes itself using a tail, which allows it to more easily control itself in three dimensions, and even in mid-air.
“We now have Salto jumping off of walls as a single maneuver and jumping on flat ground repeatedly,” Justin Yim, a fellow researcher on the project, told us. “I’m excited to work on improving Salto’s control to get it jumping repeatedly off of obstacles like walls and developing better landing control.”
A paper describing the work has been accepted for the 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS).
