R.D. Fellman

About

R.D. Fellman’s research centers on adaptive control and reinforcement learning for dynamic robotic systems, with a focus on bridging the gap between controlled laboratory environments and unpredictable real-world applications. Their foundational 1996 paper, “Reinforcement learning for dynamic robotic systems,” established adaptive algorithms as a critical solution for robotic control in unknown or changing environments, earning 3 citations for its pioneering exploration of three feedback methods. Fellman further advanced the field with their 2002 work on an efficient adaptive input quantizer for resetable dynamic robotic systems, which addressed the challenge of autonomous reactive controllers that rely solely on environmental failure signals for adaptation. This 2-citation paper demonstrated how reinforcement algorithms must simultaneously learn long-term discounted reinforcement functions across state spaces while searching for optimal control policies. Though their citation counts are modest, Fellman’s contributions represent early, foundational thinking in adaptive robotics—exploring how machines can learn from sparse feedback in real-time. Their work anticipated modern developments in reinforcement learning for robotics, particularly in systems requiring reset capabilities and efficient quantization. Fellman’s research remains relevant for students and engineers exploring autonomous control in uncertain environments.

Research Focus

Key Achievements

Top Papers

1. 1
   Reinforcement learning for dynamic robotic systems

   3 citations · 1996
2. 2
   An efficient adaptive input quantizer for resetable dynamic robotic systems

   2 citations · 2002

Key Collaborators

Contact & Links