Ruiqi Xian
Papers
5
Total Citations
29
H-Index
3
About
Ruiqi Xian is a researcher working at the intersection of computer vision, unmanned aerial vehicles (UAVs), and embodied AI, with particular expertise in aerial video understanding and vision-language model applications. His most recognized contribution is AZTR (Aerial Video Action Recognition with Auto Zoom and Temporal Reasoning), a novel framework designed for human action recognition in UAV-captured footage that operates efficiently on edge and mobile devices — a work that has garnered 19 citations since its 2023 publication. Building on this foundation, Xian has advanced the field through complementary innovations including synthetic data augmentation to address the scarcity of labeled UAV datasets and Soft Conditional Prompt Learning (SCP), which adapts large vision-language models for aerial action recognition tasks. More recently, his research has expanded into robotics, where he investigates both the promise and peril of LLM/VLM-controlled systems — exposing critical vulnerabilities to input perturbations while also developing VLM-GroNav, an outdoor robot navigation algorithm that grounds vision-language reasoning in physical terrain properties. Collectively, his work bridges the gap between aerial perception and intelligent autonomous systems, making meaningful contributions to real-world, resource-constrained deployment scenarios.
Research Focus
Key Achievements
Top Papers
- 1AZTR: Aerial Video Action Recognition with Auto Zoom and Temporal Reasoning19 citations · 2023
- 2
- 3
- 4On the Vulnerability of LLM/VLM-Controlled Robotics2 citations · 2024
- 5