Papers
2
Total Citations
10
H-Index
2
About
Yu Hua is a researcher advancing the frontiers of embodied AI and multi-agent systems. His primary research areas include 3D human motion prediction, human-robot interaction, and multi-agent reinforcement learning. Hua's most notable contribution is the development of a Deformable Transformer-based Adversarial Network for efficient 3D human motion prediction (2022, 8 citations). This work addresses two critical challenges in the field: the mode collapse problem that leads to non-plausible poses and the quadratic computational complexity of traditional transformer architectures. By introducing deformable attention mechanisms and adversarial training, Hua's approach enables more accurate and computationally tractable predictions of human motion, a fundamental requirement for seamless human-robot collaboration. Additionally, Hua has explored bio-inspired solutions for multi-agent systems, proposing a Pheromone Based Independent Reinforcement Learning framework for multi-agent navigation (2021, 2 citations). This work draws inspiration from insect swarm behavior to enable decentralized coordination among autonomous agents. Hua's research sits at the intersection of computer vision, robotics, and reinforcement learning, with potential applications in autonomous driving, assistive robotics, and virtual reality.
Research Focus
Key Achievements
Top Papers
- 1
- 2