A fully differentiable solver for simulating coupled fluid-robot dynamics

An open source visual-inertial-leg odometry with high estimation accuracy

Novel hardware design to enhance quadruped robots

Building new algorithms for learning dynamics models that are sample efficient and generalizable.

Smooth and differentiable collision detection between pairs of convex primitives, enabled by differentiable convex optimization.

Trajectory optimization with conic and complementarity constraints–plus differentiable!

Differentiable rigid-body-dynamics with contact simulator for robotic systems.

Bi-level trajectory optimization with dynamics represented as constrained optimization problems.

Fast model-predictive control for systems that make and break contact with objects and the environment.

Directly optimizing robust policies for feedback motion planning.

Real-time simulation and trajectory optimization for soft robots.

Developing a fast and robust solver for constrained dynamic games aimed at identifying Nash equilibrium strategies.

The CR3BP is a useful model for designing and analyzing spacecraft trajectories that pass between multiple large bodies. We use optimization techniques to find trajectories that meet mission constraints while being dynamically feasible in the CR3BP.

Extending lifetimes of commercial microelectronic devices in harsh radiation environments without additional shielding or device alterations.

Building new solvers for trajectory optimization problems that are fast, accurate, and numerically robust.

Designing a new, lost-cost quadrupeds with state-of-the-art control.

Optimizing long duration spacecraft maneuvers for electric propulsion.

Developing algorithms and hardware for underactuated control of small satellites, mainly through trajectory optimization techniques of magnetorquer attitude manipulation.

Scalable Cooperative Transport of Cable-Suspended Loads with UAVs using Distributed Trajectory Optimization

Developing a fast, low memory footprint algorithm to solve minimum-fuel problems with possible implementation onboard a CubeSat for embedded trajectory optimization.

An open-source, radiation-tested reliable cubesat framework programmable entirely in python.

Controlling systems that make and break contact with objects and the environment. Applications to robotic locomotion and manipulation.

Making things get where they’re supposed to go when we don’t know exactly how they move and what disturbance forces might be pushing on them.

Tiny low-cost satellites made on printed circuit boards