Kaiyuan Tan

Papers

Total Citations

H-Index

About

Kaiyuan Tan is a rising researcher at the intersection of robotics, formal methods, and artificial intelligence, with a core focus on enabling autonomous agents to understand and execute complex, temporally structured missions. Tan’s major contribution lies in bridging the gap between natural language instructions and formal task specifications, particularly through the innovative use of Large Language Models (LLMs). In their highly cited work, *"Conformal Temporal Logic Planning using Large Language Models"* (2023, 5 citations), Tan introduced a novel framework that translates natural language sub-tasks into Linear Temporal Logic (LTL) predicates, allowing mobile robots to plan and execute missions with precise temporal and logical ordering. This approach leverages conformal prediction to provide statistical guarantees on task satisfaction, marking a significant step toward trustworthy autonomous planning. Tan further advanced the field with *"Mission-driven Exploration for Accelerated Deep Reinforcement Learning with Temporal Logic Task Specifications"* (2023, 2 citations), where they developed a DRL algorithm that uses mission-driven exploration to efficiently learn control policies for agents with unknown stochastic dynamics, maximizing the probability of satisfying LTL objectives. This work addresses a critical bottleneck in reinforcement learning—sample efficiency—by guiding exploration toward temporally meaningful states. With a growing citation record and a clear trajectory toward integrating LLMs with formal verification, Kaiyuan Tan is establishing themselves as a key innovator in safe, language-guided autonomous systems.