Papers
7
Total Citations
581
H-Index
6
About
Fangchang Ma is a computer vision and robotics researcher whose work centers on depth sensing, depth completion, and resource-constrained robotic perception. His most influential contribution, "Self-Supervised Sparse-to-Dense" (2019), tackles the challenging problem of reconstructing dense depth maps from sparse LiDAR measurements fused with monocular camera imagery — a critical capability for autonomous driving and mobile robotics. With over 470 citations, this work has become a landmark reference in the depth completion field, addressing key difficulties such as irregular sparsity patterns and the absence of dense ground-truth supervision. Ma's broader research agenda explores how robots with limited computational and power budgets can still achieve reliable environmental understanding. His work on sparse depth sensing for resource-constrained robots demonstrates that meaningful 3D geometry reconstruction is achievable even with severely limited sensor data, a practically important insight for real-world deployments. His 2019 "FastDepth" paper further advances efficient monocular depth estimation tailored for embedded systems, bridging the gap between deep learning accuracy and on-device feasibility. Collectively, Ma's contributions have meaningfully shaped how the robotics and autonomous systems communities approach perception under hardware constraints.
Research Focus
Key Achievements
Top Papers
- 1
- 2Sparse sensing for resource-constrained depth reconstruction35 citations · 2016
- 3
- 4FastDepth: Fast Monocular Depth Estimation on Embedded Systems22 citations · 2019
- 5Sparse depth sensing for resource-constrained robots20 citations · 2019
- 6Sparse Depth Sensing for Resource-Constrained Robots7 citations · 2017
- 7