Taylor A. Howell
Papers
13
Total Citations
474
H-Index
8
About
Taylor A. Howell is a leading researcher at the intersection of robotics, trajectory optimization, and differentiable simulation. His work focuses on enabling robots to make and break contact with their environments—a fundamental challenge for tasks like manipulation and locomotion. Howell’s major contributions include developing ALTRO, a fast solver for constrained trajectory optimization (177 citations), and pioneering contact-implicit model predictive control (CI-MPC), which generalizes linear MPC to contact-rich settings (81 citations). He has also advanced scalable multi-robot systems, as demonstrated in his work on cooperative transport of cable-suspended loads with UAVs (82 citations). Howell’s innovations in differentiable collision detection (DCOL, 46 citations) and differentiable physics engines like Dojo (10 citations) have opened new avenues for gradient-based optimization in robotics. His open-source framework MuJoCo MPC (MJPC) enables real-time predictive control, making sophisticated planning accessible to the broader robotics community. With over 450 total citations and a portfolio spanning fast solvers, differentiable physics, and multi-agent coordination, Howell’s work is shaping the future of autonomous systems that must physically interact with the world.
Research Focus
Key Achievements
Top Papers
- 1ALTRO: A Fast Solver for Constrained Trajectory Optimization177 citations · 2019
- 2
- 3Fast Contact-Implicit Model Predictive Control81 citations · 2024
- 4Differentiable Collision Detection for a Set of Convex Primitives46 citations · 2023
- 5Differentiable Physics Simulation of Dynamics-Augmented Neural Objects30 citations · 2023
- 6Predictive Sampling: Real-time Behaviour Synthesis with MuJoCo12 citations · 2022
- 7
- 8Dojo: A Differentiable Physics Engine for Robotics10 citations · 2022
- 9Fast Contact-Implicit Model-Predictive Control8 citations · 2021
- 10Linear Contact-Implicit Model-Predictive Control.5 citations · 2021