2023-2024 Seminar Series
Upcoming Seminar
Speaker: Toshiyuki Ohtsuka (Kyoto University)
Title: Numerical Optimization for Nonlinear Model Predictive Control:From Smooth Nonlinearities to Switches, Jumps, and Logic
Abstract: Advances in computers and numerical solution methods have been expanding the application of model predictive control (MPC), which solves optimal control problems in real time to perform feedback control. In particular, nonlinear MPC (NMPC) has been gaining attention due to its applications to complex nonlinear mechanical systems. Additionally, software tools for NMPC have rapidly developed to facilitate its implementation. In this talk, I will first present a comprehensive overview of real-time optimization algorithms, software tools, and applications of NMPC. One of the key ideas there is exploiting the smoothness of nonlinearities in problems. Then, I will present recent advances in parallel computation methods for NMPC and an application of NMPC for whole-body control of a quadruped robot with switches and state jumps. Furthermore, I will also introduce ongoing work on computation methods for more general optimal control problems involving equilibrium constraints or signal temporal logic specifications, which appear in various problems in robotics.
Date: Friday, April 19th 2024
Time: 09:00-10:00 AM EDT
Link: https://columbiauniversity.zoom.us/j/91247893326?pwd=L2JWU21aQzc4cU1ZQklEb0QrWGQvdz09
Line-Up and Video Links
| Date | Speaker | Title | YouTube |
|---|---|---|---|
| Dec. 1, 2023 | Yuval Tassa | MJPC: Asynchronous UI for real-time behavior synthesis | (link) |
| Jan. 31, 2024 | Alessandro Saccon | Impact-Aware Manipulation: Exploiting Intentional Collisions to Speed Up Task Execution | (link) |
| Feb. 23, 2024 | Noémie Jaquier | From Data Structure, Physics, and Human Knowledge: A Manifold of Robotic Geometries | (link) |
| Mar. 14, 2024 | Ludovic Righetti | Learning complex behaviors with nonlinear MPC | (link) |
| Apr. 19, 2024 | Toshiyuki Ohtsuka | Numerical Optimization for Nonlinear Model Predictive Control:From Smooth Nonlinearities to Switches, Jumps, and Logic |