Papers
8
Total Citations
415
H-Index
6
About
Tim Salzmann is a researcher at the intersection of autonomous systems, human motion forecasting, and robot learning, whose work addresses fundamental challenges in making robots safer and more intelligent in human-shared environments. He is perhaps best known for developing **Trajectron++**, a multi-agent generative trajectory forecasting framework that integrates heterogeneous data to predict human motion in socially-aware robotic navigation — a contribution that has garnered over 150 combined citations and become a reference point in the self-driving and human-robot interaction communities. His work on **Real-Time Neural MPC** (191 citations) bridges deep learning with Model Predictive Control, enabling agile robotic platforms like quadrotors to leverage accurate, learned dynamics models while maintaining embedded real-time performance. Salzmann has also advanced full-body human motion forecasting through **Motron** and explored how human pose estimation can enrich trajectory prediction in dynamic indoor environments. His most recent work ventures into probabilistic modeling on SO(3) manifolds using normalizing flows, reflecting a growing interest in geometrically principled machine learning. Across his portfolio, Salzmann consistently pushes toward robots that can reason about and safely coexist with humans in complex, unstructured settings.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4Motron: Multimodal Probabilistic Human Motion Forecasting38 citations · 2022
- 5Robots That Can See: Leveraging Human Pose for Trajectory Prediction25 citations · 2023
- 6
- 7
- 8