Jiwei Shen
Papers
4
Total Citations
48
H-Index
3
About
Jiwei Shen is a rising star in embodied AI and robot autonomy, whose research bridges the critical gap between static simulation and real-world cluttered environments. His primary contributions lie in interactive visual navigation and reinforcement learning, where he has pioneered methods for robots to intelligently displace obstacles—such as shoes or boxes—to clear a path toward a goal. His landmark work, “Transformer Memory for Interactive Visual Navigation in Cluttered Environments” (2023, 22 citations), introduced a novel memory architecture that enables agents to recall and reason about past interactions, achieving state-of-the-art performance in dynamic settings. Building on this, his “VME-Transformer” (2023, 19 citations) further enhanced visual memory encoding, allowing robots to navigate more efficiently through complex, object-filled spaces. Shen has also advanced geometric correspondence estimation for SLAM systems (GCENet, 2024, 5 citations) and tackled the fundamental challenge of causal confusion in RL, proposing targeted interventions to eliminate spurious correlations (2024, 2 citations). With over 48 cumulative citations in just two years, Shen’s work is shaping the next generation of truly autonomous robots that can operate in the messy, interactive world we inhabit.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4