Hadi Beik-mohammadi
Hadi Beik-mohammadi co-authored 2021 research on learning Riemannian manifolds for robot motion from demonstrations. The method uses a variational autoencoder to represent motions and geodesics to plan paths. A separate 2021 paper combines reinforcement learning with model-mediated teleoperation using a hand-arm exoskeleton. In a 2024 manuscript revised in 2025, he co-authored extensions to neural contractive dynamical systems for several tasks and obstacle avoidance.
- Works
- 23
- Citations
- 108
- h-index
- 7
Publication numbers from OpenAlex, as of 14 October 2026.
Timeline
-
2021
Learning a geometry of movement
Beik-mohammadi co-authored work on Riemannian manifolds for motion skills, first submitted in June 2021. The method learns from demonstrations using a variational autoencoder over robot end-effector positions and orientations. The learned geometric space provides a way to plan paths between different poses.
-
2021
Adapting demonstrated skills
The same paper studies geodesic motion skills and adaptation to obstacles by changing the surrounding metric. It reports experiments with a 7-degree-of-freedom manipulator. The authors also examine movements assembled from several demonstrated modes.
-
2021
Learning in teleoperation
He co-authored a separate July 2021 paper on model-mediated teleoperation with a hand-arm exoskeleton. It combines dynamic movement primitives with reinforcement learning to address environmental change and model uncertainty. The paper reports tests with DLR’s Exodex Adam system.
-
2024–2025
Neural dynamics and stability
His later co-authored manuscript extends neural contractive dynamical systems. First submitted in November 2024 and revised in September 2025, it studies regularisation, several tasks, and uncertainty-driven latent obstacle avoidance.
Sources and credits
Photo: alr.iar.kit.edu; ALR - Team; creator not stated. License: Not stated by source. Source: alr.iar.kit.edu. Cropped to a square and resized. Ask us to remove or replace this photo.