Papers
25
Total Citations
1,200
H-Index
13
About
Thomas Lampe is a robotics and machine learning researcher whose work sits at the intersection of deep reinforcement learning (RL) and real-world robotic control. His research addresses some of the field's most persistent challenges: sparse rewards, data efficiency, and the sim-to-real gap that prevents laboratory algorithms from translating into practical robotic systems. Lampe's most influential contribution, "Leveraging Demonstrations for Deep Reinforcement Learning on Robotics Problems with Sparse Rewards" (2017, 510 citations), introduced a framework that combines human demonstrations with autonomous experience to accelerate learning under sparse reward conditions — a breakthrough for practical robotics deployment. His subsequent work on Scheduled Auxiliary Control (SAC-X, 155 citations) further advanced the field by enabling agents to acquire complex behaviors from scratch using auxiliary task structures. Beyond reward shaping, Lampe has made significant contributions to offline RL, sim-to-real transfer, hybrid discrete-continuous control, and even brain-computer interface-driven robotic systems — demonstrating a remarkably broad research vision. His early work on visual servoing with neural reinforcement learning (2013) reflects a career-long commitment to grounding algorithmic advances in genuine robotic applicability. With hundreds of cumulative citations, Lampe stands as a meaningful contributor to making reinforcement learning viable for real-world manipulation tasks.
Research Focus
Key Achievements
Top Papers
- 1
- 2Learning by Playing - Solving Sparse Reward Tasks from Scratch155 citations · 2018
- 3Data-efficient Deep Reinforcement Learning for Dexterous Manipulation118 citations · 2017
- 4
- 5Self-Supervised Sim-to-Real Adaptation for Visual Robotic Manipulation56 citations · 2020
- 6
- 7
- 8
- 9Continuous-Discrete Reinforcement Learning for Hybrid Control in Robotics27 citations · 2020
- 10