Research
Research statement
My research training focused on contact-rich manipulation when geometric uncertainty exceeds what position control alone can guarantee. Peg-in-hole assembly with tight clearance is a canonical benchmark: it requires combining perception (where is the hole?), control (how hard to push?), and human-in-the-loop strategies when autonomy is brittle.
At Innopolis University I studied this on a KUKA LBR iiwa with ROS, a 3D Systems Touch haptic device, and VR-based operator interfaces. My thesis compared four control modes: full autonomy, bilateral teleoperation with force feedback, shared autonomy, and learning from demonstration — with user studies across scenarios. That line of work extended hybrid vision/force control and deep-learning-based picking, published at IEEE NIR 2021.
Interests
- Robot manipulation, teleoperation, and haptic interfaces
- Computer vision for detection and state estimation in cluttered scenes
- Reinforcement learning and imitation learning for contact tasks
- Human–robot interaction and shared autonomy