Bryan Chen
Papers
2
Total Citations
21
H-Index
2
About
Bryan Chen’s research lies at the intersection of computer vision, robotics, and cognitive AI, with a focus on building robust, perceptually-aware autonomous systems. His most cited work, “Robust Policies via Mid-Level Visual Representations” (2020, 19 citations), challenges the prevailing end-to-end deep reinforcement learning paradigm by demonstrating that decoupling perception from control and using mid-level visual features yields policies that are both sample-efficient and resilient to environmental changes—a critical insight for real-world manipulation and navigation tasks. More recently, in “Out of Sight, Still in Mind” (2024), Chen tackles the fundamental problem of object permanence in robotics. He introduces DOOM and LOOM, novel memory-augmented frameworks that enable robots to reason and plan around occluded objects by integrating video-based tracking with object-oriented memory models. This work pushes multi-object manipulation toward greater reliability in cluttered, dynamic environments. Though early in his career, Chen’s contributions are already shaping how the field thinks about the trade-offs between end-to-end learning and modular design, and his work on memory-driven planning points toward a future where robots operate with a deeper, more human-like understanding of their surroundings.
Research Focus
Key Achievements
Top Papers
- 1
- 2