Karina Palyutina
Papers
1
Total Citations
6
H-Index
1
About
Karina Palyutina is a robotics researcher whose work focuses on bridging the gap between reinforcement learning (RL) and real-world autonomous navigation. Her primary research areas include collision avoidance, local path planning, and the application of deep RL algorithms to physical robotic systems. Her most notable contribution is the development of SACPlanner, a Soft Actor Critic-based local planner that leverages polar state representations to achieve robust, real-world collision avoidance. In her highly cited 2023 paper, Palyutina demonstrated that recent enhancements to the SAC algorithm—such as RAD and DrQ—enable near-perfect training performance after only 10,000 episodes, a significant milestone for deploying RL in practical robotics. This work has garnered 6 citations and stands out for its rigorous evaluation of training efficiency and trajectory quality on actual robots, rather than in simulation alone. By systematically comparing RL-based planners with traditional approaches, Palyutina has provided a clear pathway for integrating learning-based methods into ROS-based navigation stacks. Her research is essential reading for students and engineers seeking to understand how to make RL reliable and sample-efficient for real-world autonomous systems.
Research Focus
Key Achievements
Top Papers
- 1