Papers
6
Total Citations
129
H-Index
5
About
Brendan Michael is a robotics researcher whose work spans continuum robot dynamics, imitation learning, and data-driven control for soft robotic systems. His most influential contribution, the *TMTDyn* MATLAB package (2020, 86 citations), established a practical framework for modeling and controlling hybrid rigid–continuum robots using discretized lumped systems and reduced-order models — addressing a critical need for fast, accurate, and accessible dynamic models within the robotics community. This work has become a widely adopted resource for researchers navigating the complex design and control challenges inherent to continuum robotics. Beyond dynamic modeling, Michael has made notable strides in robust imitation learning, developing Bayesian Disturbance Injection techniques that enable robots to learn flexible, generalizable policies even when demonstrations are diverse or suboptimal — a persistent challenge in real-world deployment. His exploration of Deep Koopman methods for soft robot control further reflects his commitment to bridging machine learning with physics-informed modeling, offering spectral analysis tools that sidestep expensive analytical formulations. Early work on task-parameterised movement learning rounds out a research profile defined by a drive to make robot learning more adaptable, reliable, and practically deployable across complex manipulation scenarios.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3Deep Koopman with Control: Spectral Analysis of Soft Robot Dynamics11 citations · 2022
- 4Disturbance-injected Robust Imitation Learning with Task Achievement9 citations · 2022
- 5
- 6