Laszlo-Peter Berczi
Papers
4
Total Citations
56
H-Index
4
About
Laszlo-Peter Berczi is a roboticist whose work focuses on enabling autonomous navigation in unstructured, outdoor environments. His primary research areas include terrain assessment, visual teach-and-repeat (VT&R) navigation, and learning from human demonstration. Berczi’s major contribution is the development of place-dependent, learning-based systems that allow robots to safely repeat previously driven paths by intelligently assessing terrain. His 2015 paper, “Learning to assess terrain from human demonstration using an introspective Gaussian-process classifier” (22 citations), pioneered a method where a robot learns to distinguish traversable from untraversable terrain after a short period of human supervision. This work was extended in his 2016 paper, “It’s like Déjà Vu all over again” (8 citations), which introduced a classifier that improves over time by exploiting the repetitive nature of VT&R tasks. His 2017 field test, “I Can See for Miles and Miles” (22 citations), validated the robustness of VT&R 2.0 over long distances. Berczi’s research is notable for its practical, data-driven approach to a critical challenge in field robotics, bridging the gap between supervised learning and autonomous long-term operation.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4