Tim Harley
Papers
3
Total Citations
107
H-Index
3
About
Tim Harley is a researcher working at the intersection of artificial intelligence, natural language processing, and embodied agent design. His work focuses on building interactive AI agents capable of understanding and following human instructions in simulated environments — a foundational step toward the intelligent robots long envisioned in science fiction. Harley's most influential contribution, "Imitating Interactive Intelligence" (2020, 43 citations), explores how artificial agents can interact naturally with humans through language and perception, combining imitation learning with rich multimodal inputs. This line of inquiry extended into "Creating Multimodal Interactive Agents with Imitation and Self-Supervised Learning" (2021, 32 citations), which advances agent design using self-supervised techniques to improve generalization and interactivity. His research also tackles the challenge of bridging language and motor control, as demonstrated in "Human Instruction-Following with Deep Reinforcement Learning via Transfer-Learning from Text" (2020, 32 citations), where he investigates how reinforcement learning agents can be made more responsive to real human commands through text-based transfer learning. Collectively, Harley's work has garnered over 100 citations and represents meaningful progress toward AI systems that can collaborate with humans in physical and simulated worlds.
Research Focus
Key Achievements
Top Papers
- 1Imitating Interactive Intelligence43 citations · 2020
- 2
- 3