Papers
2
Total Citations
34
H-Index
2
About
Dongseok Im is a leading researcher in mobile spatial computing, specializing in energy-efficient hardware accelerators for real-time 3D perception and augmented reality (AR) systems. His work centers on developing specialized processors that enable autonomous robots and AR glasses to perform complex spatial tasks—such as simultaneous localization and mapping (SLAM) and dense depth sensing—within strict power and latency constraints. Im’s major contributions include the "Space-Mate" processor, a 303.5mW real-time sparse mixture-of-experts-based NeRF-SLAM accelerator (2024, 18 citations), which achieves unprecedented efficiency for mobile spatial computing by combining neural radiance fields with SLAM. He also designed the "DSPU," a 281.6mW depth signal processing unit (2022, 16 citations) that performs deep learning-based dense RGB-D data acquisition, depth fusion, and 3D bounding box extraction at over 30 fps. These innovations have been recognized for pushing the boundaries of on-device AI, enabling real-time cyber-physical interaction without cloud dependency. Im’s work is pivotal for next-generation mobile platforms, bridging the gap between high-accuracy 3D perception and ultra-low-power deployment.
Research Focus
Key Achievements
Top Papers
- 1
- 2