Patrick van der Smagt
Patrick van der Smagt’s work includes spatial world models, robot trajectories and language-informed visual models. These projects connect probabilistic learning with robot control. He coauthored PRISM, a CoRL 2023 study of probabilistic real-time inference in spatial world models. Van der Smagt co-authored a paper that learns demonstrated trajectories with a conditional variational model and autoregressive decoding. The method adapts paths to obstacles through guided decoding and beam search, with robot-platform experiments. He coauthored the ICRA 2025 paper LIMT on language-informed multi-task visual world models with Elie Aljalbout, Nikolaos Sotirakis, Maximilian Karl and Nutan Chen.
- Now at
- Volkswagen Group
- Works
- 259
- Citations
- 13,345
- h-index
- 40
Publication numbers from OpenAlex, as of 14 October 2026.
Timeline
-
2023
Spatial world models
He coauthored PRISM, a CoRL 2023 study of probabilistic real-time inference in spatial world models.
-
2024
Guided robot-path decoding
Van der Smagt co-authored a paper that learns demonstrated trajectories with a conditional variational model and autoregressive decoding. The method adapts paths to obstacles through guided decoding and beam search, with robot-platform experiments.
-
2025
Language-informed visual models
He coauthored the ICRA 2025 paper LIMT on language-informed multi-task visual world models with Elie Aljalbout, Nikolaos Sotirakis, Maximilian Karl and Nutan Chen.
Sources and credits
Photo: eliza.school; Prof. Dr. Patrick van der Smagt - Zuse School ELIZA; creator not stated. License: Not stated by source. Source: eliza.school. Cropped to a square and resized. Ask us to remove or replace this photo.