Olivia Watkins

Berkeley College, University of California, Berkeley

Papers

Total Citations

H-Index

About

Olivia Watkins is a rising star in artificial intelligence and robotics, whose research bridges the critical gap between simulated training and real-world deployment. Her most impactful work, "Auto-Tuned Sim-to-Real Transfer" (2021, 46 citations), addresses the notorious "reality gap" by introducing a method that automatically tunes simulation parameters to match real-world dynamics, eliminating the need for labor-intensive manual domain randomization. This innovation has significant implications for scaling robot learning efficiently. Watkins also pioneers the integration of large language models with reinforcement learning, as demonstrated in her 2023 paper (39 citations), which uses LLMs to guide pretraining and exploration in environments lacking dense reward functions—a breakthrough for tackling complex, sparse-reward tasks. Her commitment to human-robot interaction is evident in her work on explaining robot policies (2021), where she proposes using illustrative examples to build accurate mental models for users, enhancing trust and safety. With a total of 95 citations across her key papers, Watkins is establishing herself as a leading voice in sim-to-real transfer and foundation-model-guided RL, shaping how robots learn and interact in the real world.