Papers
49
Total Citations
1,249
H-Index
22
About
Ingmar Posmer is a leading researcher in robotics and machine perception, with key contributions in deep learning for tracking, scene understanding, and robust autonomous navigation. His work on end-to-end object tracking, notably "Deep Tracking" (180 citations), pioneered direct mapping from raw sensor data to object tracks using recurrent neural networks, eliminating the need for hand-crafted features. He also advanced object-centric generative models with the GENESIS framework (over 145 citations combined), enabling compositional scene inference critical for robotics and reinforcement learning. Posmer has addressed real-world challenges like appearance change in outdoor robotics through adversarial domain adaptation (66 citations) and developed introspective classification methods (42+ citations) that allow robots to know when they don't know—a vital capability for mission-critical decision-making. His research on efficient lidar-based localization (38 citations) and probabilistic prediction of perception performance (37 citations) further demonstrates his impact on reliable, large-scale autonomous systems. With a focus on bridging deep learning and practical robotics, Posmer's work has shaped how robots perceive, track, and reason in dynamic, partially observable environments.
Research Focus
Key Achievements
Top Papers
- 1Deep Tracking: Seeing Beyond Seeing Using Recurrent Neural Networks180 citations · 2016
- 2
- 3
- 4
- 5
- 6
- 7Introspective classification for robot perception42 citations · 2015
- 8
- 9
- 10