Lab Meeting Schedule

May 14th: Contact Robust Trajectory Optimization
Presented by Luke Drnach, in preparation for ICRA 2021

May 21st: Group meeting for summer planning

May 28th: Adverserial training for manipulation planning
Presented by Zhigen Zhao

June 4th: Reactive synthesis for crossed leg walking under exogenous disturbance​
Presented by Nathan Boyd and Zhaoyuan Gu

June 11th: Terrain perception for quadruped locomotion
Presented by Danny Lee

June 18th: Metric temporal logic for object-oriented manipulation contact planning
Presented by Yuki Yoshinaga

June 25th: Contact Estimation via Inverse Optimization
Presented by Luke Drnach

July 2nd: No meeting

July 9th: Runtime Enforcement and Resythesis (canceled)
Presented by Jonas Warnke

July 16th: Quadruped Agility and Learning Robust Composition
Presented by Aziz Shamsah

July 23rd: Gelsight touch sensing and estimation for manipulation
Presented by Yunhai Han

July 30th: Perception for human-robot interaction
Presented by Michael Park

August 6th: Multi-agent task and motion planning
Presented by Ziyi Zhou

August 13rd: Summer Research Summary
Presented by the LIDAR members

January 22: Differentiable Optimization and Policy Learning for Manipulation
Presented by Zhigen Zhao

January 29: Current Trends in Robot Locomotion Research
Presented by Nathan Boyd, Zhaoyuan Gu, and Aziz Shamsah [Presentation Slides]

February 5: Sim2Real Tricks on Cassie and RL+Control Thoughts for Quadrupeds
Presented by Dennis Da from Nvidia

Working on the legged robot has a stiff learning curve. There is an incredible amount of theoretical proof that can distract newcomers. Once the fascinating math has been applied in the simulation, the sim2real gap could take years to overcome. Since many students, including me, have gone through this unpleasant process, I hope my sharing can flatten the learning curve and make the legged robot easier to start with. Meanwhile, the classic hybrid control theory and nonlinear trajectory optimization have limits, especially on real-time computing. We might use reinforcement learning to share some computing offline. I am going to present my recent work on Laikago in Nvidia. I have done control via model-based and RL separately and combined.

February 12: Robust Legged Robot State Estimation Using Factor Graph Optimization
Presented by Danny (Dong Jae) Lee [Presentation Slides]

February 19: Learning Differentiable Soft Contact Models, Signal Temporal Logic and Mixed Integer Programming
Presented by Michael Park and Yuki Yoshinaga [Presentation Slides]

February 26: Lab Publication Planning and Discussion
Presented by all LIDAR members

March 5: Current Trends in Trajectory Optimization
Presented by Luke Drnach and Ziyi Zhou [Presentation Slides]

March 12: Robust walking based on MPC with viability-based feasibility guarantees
Presented by Zhaoyuan Gu and Nathan Boyd

March 19: Task and Motion Planning for Safe Locomotion Navigation in Partially Observable Environments
Presented by Aziz Shamsah and Jonas Warnke

March 26: Trajectory Optimization and Learning Deformable Contact Modeling for Manipulation
Presented by Zhigen Zhao and Michael Park

April 2: Closing the Loop between Trajectory Optimization and Contact Model Estimation
Presented by Luke Drnach

April 9: Mini Cheetah Overview and Research Update
Presented by Ziyi Zhou

April 16: SLAM and TAMP for A1 Quadruped
Presented by Danny (Dong Jae) Lee

April 23: Formal Methods and Temporal Logic
Presented by Jonas Warnke and Yuki Yoshinaga

August 27: Locomotion Planning through a Hybrid Bayesian Trajectory Optimization
Presented by Luke Drnach

September 3: Controlling Physics-Based Characters Using Soft Contacts
Presented by Michael Park

September 10: Efficient Symbolic Reactive Synthesis for Finite-Horizon Tasks
Presented by Yuki Yoshinaga

September 17: Contact-Implicit Trajectory Optimization using an Analytically Solvable Contact Model for Locomotion on Variable Ground
Presented by Ziyi Zhou

September 24: Composition of Templates for Transitional Pedipulation Behaviors
Presented by Aziz Shamsah

October 1: Iterative Temporal Motion Planning for Hybrid Systems in Partially Unknown Environments
Presented by Jonas Warnke

October 8: Finding Locomanipulation Plans Quickly in the Locomotion Constrained Manifold
Presented by Jialin Li

October 15: Optimal Reduced Order Modeling of Bipedal Locomotion
Presented by Zhaoyuan Gu

October 22: Informing Multi-Modal Planning with Synergistic Discrete Leads
Presented by Zhigen Zhao

October 29: Approximate Simulation for Template Based Whole-Body Control
Presented by Nathan Boyd