Please check out a full list of videos on our lab youtube channel.

Check out this fresh new LIDAR video and enjoy a virtual lab tour with us!

———————————————————————————————————————-

The PI gave a seminar at USC Center for Cyber-Physical Systems and the Internet of Things on Robust Planning and Decision-making for Safe Legged Locomotion. Please check it out to learn about our recent works on formal methods for locomotion and trajectory optimization.

———————————————————————————————————————-

Buzzy Cassie robust walking over perturbed woodblocks. A preliminary version of our reduced-order phase-space planning method is implemented [safe-loco-2020]. We are actively developing safe terrain-aware locomotion strategies for more complex navigation tasks. Stay tuned!

———————————————————————————————————————-

Our latest submission from Zhigen Zhao, Ziyi Zhou, and Michael Park on Symbolic-Decision-Embedded Bilevel Optimization for Long-Horizon Manipulation in Dynamic Environments. This bilevel optimization is formulated as a Task and Motion Planning (TAMP) composed of causal graph task decomposition, PDDL symbolic planning, and ADMM-based trajectory optimization. The open-source code is available.

———————————————————————————————————————-

A recent submission from Luke Drnach on robust contact-implicit trajectory optimization over uncertain terrain with stochastic complementarity. Uncertainty stemming from the terrain is modeled and used to design corresponding risk-sensitive objectives under the framework of contact-implicit trajectory optimization. Uncertainties from the terrain contact distance and friction coefficients are parameterized to propose a corresponding expected residual minimization cost design approach. The open-source code is here.

———————————————————————————————————————-

Jonas Warnke and Aziz Shamsah present their joint work on Safe Locomotion Navigation in Partially Observable Environments at CDC 2020 (virtual). Here are the CDC version and open-source code. We currently target generalized locomotion tasks and experimental implementations in more complex environments.

———————————————————————————————————————-

Ziyi Zhou delivered his ACC talk on distributed trajectory optimization for legged locomotion using the alternating direction method of multipliers (ADMM). Our current work aims for generalizing this approach to more complex legged systems such as Cassie and solve the optimal foot contact sequence in this distributed optimization.

———————————————————————————————————————-

Ye gave a talk at Dynamic Walking on the robust decision-making and planning of agile locomotion in constrained environments. Thank DW organizers for the invitation.

———————————————————————————————————————-

This video presented our recent work on Cassie’s foot soft suite sensing and terrain classification algorithms. This is a collaboration with Dr. Frank Hammond’s lab. The result was published in RoboSoft 2020.