Michael Przystupa
Papers
2
Total Citations
8
H-Index
2
About
Michael Przystupa is a roboticist advancing the frontier of dexterous manipulation through data-driven control and movement representation. His research centers on learning low-dimensional action spaces and probabilistic movement primitives to make robotic systems more sample-efficient and generalizable. In his 2023 work on state-conditioned linear mappings, Przystupa demonstrated that learning a linear mapping from a low-dimensional latent space to motor commands can balance the expressiveness of nonlinear methods with the simplicity and tractability of linear control—a key insight for scaling manipulation policies. His concurrent paper on deep probabilistic movement primitives introduced a Bayesian aggregator to fuse multiple demonstrations, enabling robots to reproduce complex movements with temporal modulation while quantifying uncertainty. Though early in his career, each of these papers has already garnered 4 citations, signaling growing interest from the manipulation and imitation learning communities. Przystupa’s contributions are particularly notable for their focus on principled, interpretable architectures that retain the sample efficiency of classical robotics while leveraging modern deep learning—a promising direction for real-world robotic applications where data is scarce and reliability is paramount.
Research Focus
Key Achievements
Top Papers
- 1
- 2Deep Probabilistic Movement Primitives with a Bayesian Aggregator4 citations · 2023