Fuxiao Liu

Papers

Total Citations

H-Index

About

Fuxiao Liu is a researcher at the forefront of artificial intelligence, with a focus on the intersection of computer vision, natural language processing, and robotics. His work is distinguished by its emphasis on enhancing model robustness and adaptability in complex, real-world environments. Liu’s key contributions include the development of **Soft Conditional Prompt Learning (SCP)** for aerial video action recognition, a novel approach that leverages prompt learning to improve the accuracy of action prediction for individual agents in drone-captured footage. This work, published in 2024, has already garnered 3 citations, signaling its early impact in the field of video understanding. Additionally, Liu has made critical contributions to the safety and reliability of AI systems through his study on the vulnerability of LLM/VLM-controlled robotics. This research, also from 2024 and with 2 citations, exposes how slight input variations can compromise the performance of robots integrated with large language and vision-language models, highlighting a crucial area for future development. By addressing both performance and security, Liu is shaping a more trustworthy and capable generation of AI-driven systems.