Quadrupedal Locomotion

Legged robots have the potential to go where traditional robotic platforms that use wheels or treads cannot. This level of mobility is critical for robots that are developed to work as part of a human team or squad. Beyond just the mobility aspect, though, the robots must be capable of both maintaining pace with their human counterparts, and functioning for an extended period of time. If possible, these robots can then be used for countless tasks such as reconnaissance and load carrying, assisting their human team members as they go about their work.

As part of the Army Research Lab's (ARL) Robotics Collaborative Technology Alliance (RCTA), we are developing new planning and control algorithms for quadrupedal robots. The goal of our project is to equip the robot LLAMA, developed by NASA JPL, with the skills it needs to move at operational tempo over difficult terrain to keep up with a human squad. This requires innovative perception, planning, and control techniques to make the robot both precise in execution for navigating technical obstacles and robust enough to reject disturbances and recover from unknown errors.

As part of this work, we are extending  some of our work developed for the learning locomotion project and our humanoids on footstep planning to develop new techniques for determing both body path and footstep stequences for navigating rough terrain. We are also researching gait design for determing high-speed gaits such as galloping, which should be possible with this new hardware. Lastly, we are particularly interested in reactive strategies to maintain stability that consider complex terrain elements such as non-convex and compliant surfaces.