Yujie Cai

Papers

Total Citations

H-Index

About

Yujie Cai is a robotics researcher whose work bridges formal methods and deep reinforcement learning to tackle complex motion planning challenges. His key research areas include multi-task motion planning, stochastic environment navigation, and the integration of temporal logic specifications with learning-based control. Cai’s major contribution lies in developing a novel framework that combines Generalized Reactivity(1) (GR(1)) specifications with deep reinforcement learning, enabling robots to satisfy multiple task objectives while adapting to uncertain, dynamic environments. This approach addresses a critical limitation of standard DRL methods—sparse and poorly shaped rewards—by leveraging formal guarantees to guide exploration and improve sample efficiency. His most-cited paper, “GR(1)-Guided Deep Reinforcement Learning for Multi-Task Motion Planning under a Stochastic Environment” (2022), has garnered 8 citations and demonstrates how logical constraints can be embedded into learning pipelines to achieve reliable, safe behavior. This work is particularly notable for its potential to scale to real-world applications such as autonomous navigation and robotic manipulation. Cai’s research represents an important step toward verifiable, adaptive autonomy, offering a principled way to combine the flexibility of learning with the rigor of formal verification.