Ozsel Kilinc
Papers
2
Total Citations
13
H-Index
2
About
Ozsel Kilinc’s research centers on advancing robotic manipulation through reinforcement learning, with a particular focus on overcoming the challenge of sparse reward environments. In their most cited work, “Reinforcement learning for robotic manipulation using simulated locomotion demonstrations” (2021, 11 citations), Kilinc introduced a novel approach that leverages simulated locomotion demonstrations to learn complex manipulation skills, effectively bypassing the need for handcrafted reward functions. This work has been influential in demonstrating how transfer learning from locomotion tasks can accelerate the acquisition of dexterous manipulation policies. Building on this, their 2020 paper “Follow the Object: Curriculum Learning for Manipulation Tasks with Imagined Goals” (2 citations) proposed an innovative curriculum learning framework that uses imaginary object goals to guide the agent’s learning process. By first training the object of interest to reach desired states, Kilinc’s method enables more efficient exploration and skill acquisition in sparse-reward settings. These contributions represent important steps toward more sample-efficient and generalizable robotic learning systems, with potential applications in industrial automation and assistive robotics.
Research Focus
Key Achievements
Top Papers
- 1
- 2