Embedded Files
Physics-informed deep learning for sensing and control of a pneumatic soft manipulator BR2
Physics-informed deep learning for sensing and control of a pneumatic soft manipulator BR2
We proposed a physics-informed deep learning framework for the problem of estimating the 3D shape of a soft continuum arm from noisy measurements of the pose at a finite number of locations along the arm. Such framework is evaluated on a simulated octopus muscular arm and a physical BR2 pneumatic soft manipulator. The on-going work is how to use the reconstructed smooth shape as sensory feedback for control purpose.
Modeling, simulation, and control of a CyberOctopus soft continuum arm
Modeling, simulation, and control of a CyberOctopus soft continuum arm
Flexible octopus arms are excellent candidates for studying the intricate interplay between continuum mechanics and sensorimotor control. As opposed to articulated limbs in humans, octopus arms are soft and possess a complex muscular architecture that provides exquisite manipulation control.
We develop mathematical models for simulation of octopus arm system and apply control theoretical approaches, including optimal control and sensory feedback control for various of target-oriented tasks.
Reinforcement learning for bio-inspired robot locomotion
Reinforcement learning for bio-inspired robot locomotion
We constructed a bio-inspired framework using coupled oscillator feedback particle filter and continuous-time Q-learning to obtain the optimal control of periodic locomotion gaits for robotic systems modeled as coupled rigid bodies. The framework is under the partially observed settings and does not require knowledge of the explicit form of the dynamics or the observation model.
Page updated
Google Sites
Report abuse