Ioannis Exarchos
Papers
6
Total Citations
131
H-Index
5
About
Ioannis Exarchos is a leading researcher at the intersection of robotics, physics simulation, and optimal control. His primary contributions lie in differentiable physics, where he developed Nimble, a fast and feature-complete differentiable physics engine for articulated rigid bodies with contact constraints. This work, which has garnered over 80 combined citations, enables gradient-based optimization for robotic control and learning, bridging the gap between simulation and real-world hardware. Exarchos has also advanced decision-making under uncertainty by pioneering methods that connect nonlinear stochastic optimal control theory with machine learning, using forward-backward stochastic differential equations (SDEs) to learn deep stochastic optimal control policies. His research extends to soft robotics, where he has explored task-specific design optimization for inflated-beam soft robots with growable discrete joints, aiming to create safer, lower-cost manipulators for home environments. Additionally, Exarchos has addressed the sim-to-real gap through kinematic domain randomization and adaptation, facilitating the transfer of reinforcement learning policies from simulation to physical robots. His work is notable for its practical impact, offering tools and frameworks that empower researchers to design, simulate, and control complex robotic systems with unprecedented efficiency and realism.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4
- 5Policy Transfer via Kinematic Domain Randomization and Adaptation5 citations · 2021
- 6Stochastic optimal control - a forward and backward sampling approach5 citations · 2017