Dynamic Locomotion

Animals, including humans and apes, have demonstrated impressive agility in traversing obstacles by intelligently exploiting contact forces in their environment. Studying the mechanics and planning behind their dynamic forms of locomotion can yield a simpler, more elegant, and more agile style of robotic locomotion in challenging environments.

In advancing the state-of-the-art, our work currently focuses on hybrid mechanical systems that undergo impulsive effects (e.g. the dynamics of a bouncing ball). Robots that walk, hop, or brachiate fall within this class of mechanical systems. The goal of our work is to produce theoretical tools for planning, control, and estimation of these robots allowing them to move dynamically (e.g. acrobatically, aggressively, nimbly, quickly, etc.) in their environment. A major challenge is dealing with the nonsmooth equations of motion where traditional solution methods for problems in planning, control, and estimation may not apply.

 

Research Projects:

Click on a project to learn more.

Gait Generation for Underactuated Hybrid Systems

We present a method for generating gaits for multi-degree-of-freedom, underactuated robots with hybrid dynamics in a unified and extensible framework.

The Gibbot

The Gibbot is a dynamic climbing robot developed for experimental validation of estimation and control of hybrid mechanical locomotion systems.

ParkourBot

Inspired by the flexibility of parkour, Carnegie Mellon University, in collaboration with Northwestern University, developed the ParkourBot, a dynamic climbing robot equipped with two springy BowLegs.

Related publications