Kai Oliver Arras
Papers
13
Total Citations
194
H-Index
6
About
Kai Oliver Arras is a leading researcher in mobile robotics and human-robot interaction, with key contributions in robot navigation, people tracking, and social robotics. His foundational work on feature-based navigation in known and unknown environments, detailed in his 2003 EPFL thesis (57 citations), established robust methods for autonomous robot localization and mapping. Arras pioneered the design and deployment of fully autonomous tour guide robots, bridging academic research and industry through a collaboration that produced 11 machines via the spin-off BlueBotics SA. His later research focuses on human-aware robotics, including multi-model hypothesis tracking of groups using RGB-D data (35 citations) and deep person detection in 2D range data. A notable contribution is his study on the robustness of 3D human pose estimation to occlusion (50 citations), which revealed critical vulnerabilities in state-of-the-art methods for realistic environments. Arras also explored tele-existence and social interaction through the DARYL robot, investigating embodiment and operator modality. With over 180 citations across his most-cited works, his research has significantly advanced the integration of robots into human-centered spaces, from narrow labyrinthine environments to public exhibitions, making him a key figure in autonomous systems and socially interactive robotics.
Research Focus
Key Achievements
Top Papers
- 1Feature-based robot navigation in known and unknown environments57 citations · 2003
- 2How Robust is 3D Human Pose Estimation to Occlusion?50 citations · 2018
- 3Multi-model hypothesis tracking of groups of people in RGB-D data35 citations · 2014
- 4Robot Navigation in Centimeter Range Labyrinths10 citations · 2001
- 5
- 6
- 7Robots in Exhibitions: Proceedings Workshop WS96 citations · 2002
- 8Deep Person Detection in 2D Range Data5 citations · 2018
- 9
- 10Robotic tele-presence with DARYL in the wild3 citations · 2014