Amit Dekel
Papers
2
Total Citations
9
H-Index
2
About
Amit Dekel is a leading researcher in probabilistic robotics and visual localization, with a focus on enabling autonomous systems to navigate reliably in complex, ambiguous environments. His work addresses a critical challenge: when robots encounter repetitive structures or symmetrical scenes, traditional localization methods often fail, producing a single, incorrect pose estimate. Dekel’s major contributions include developing a probabilistic framework for visual localization in ambiguous scenes, which explicitly models multiple, equally likely camera poses rather than forcing a single guess. This approach, detailed in his 2023 paper (7 citations), provides a robust foundation for relocalization when robots lose track. Building on this, his 2024 work on conditional variational autoencoders for probabilistic pose regression (2 citations) introduces a deep generative model that outputs a distribution over possible poses, further enhancing resilience in repetitive environments. Though early in citation impact, these papers are foundational for next-generation autonomous navigation, offering a principled way to handle uncertainty. Dekel’s research is pivotal for field robotics, where ambiguous scenes—from warehouse aisles to urban canyons—are the norm, not the exception.
Research Focus
Key Achievements
Top Papers
- 1A Probabilistic Framework for Visual Localization in Ambiguous Scenes7 citations · 2023
- 2Conditional Variational Autoencoders for Probabilistic Pose Regression2 citations · 2024