Papers
4
Total Citations
15
H-Index
3
About
Boyu Li is an emerging researcher specializing in deep reinforcement learning, robot control, and representation learning, with a particular focus on developing generalizable and adaptive policies for complex control tasks. His work spans several innovative frontiers, including morphology-agnostic policy learning and unsupervised pretraining for visual control. Li's most notable contribution, "MAT: Morphological Adaptive Transformer" (2024, 6 citations), introduces a transformer-based framework capable of simultaneously controlling robots with diverse morphologies — a significant step toward truly universal robotic control. His earlier work on moving target shooting control (2021, 4 citations) demonstrated the practical applicability of deep reinforcement learning to dynamic robotic environments, reflecting his commitment to bridging theory and real-world application. Perhaps most ambitiously, Li has pioneered cross-domain reinforcement learning pretraining through his CRPTpro framework (2025, 3 citations), which leverages prototype-based self-supervised learning to enable efficient knowledge transfer across domains — a challenging and impactful problem in continuous visual control. With a growing citation record and contributions across multiple high-impact research directions, Li represents a promising voice in the next generation of reinforcement learning and intelligent robotics research.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4