Tristan Laidlow
Papers
3
Total Citations
130
H-Index
3
About
Tristan Laidlow is a leading researcher in visual robotic manipulation and 3D scene understanding, whose work bridges the gap between perception and action in autonomous systems. His primary contributions lie in developing efficient learning frameworks for robotic control and real-time dense reconstruction for simultaneous localization and mapping (SLAM). Laidlow’s most influential work, “Coarse-to-Fine Q-attention” (66 citations), revolutionizes continuous control by replacing unstable actor-critic methods with a discretized reinforcement learning approach, dramatically improving data efficiency for visual manipulation tasks. His earlier landmark paper, “DeepFusion” (60 citations), introduced a method for real-time dense 3D reconstruction from monocular SLAM using single-view depth and gradient predictions, overcoming the limitations of depth cameras and enabling robust outdoor operation. More recently, Laidlow’s “SIMstack” work tackles the challenging problem of estimating 3D shape and instance information from single views of unordered object stacks, advancing scene understanding for cluttered environments. His research consistently addresses practical robotics challenges—from efficient learning to spatial perception—making his methods widely adopted in both academic and applied settings.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3