Mannes Poel
Papers
16
Total Citations
243
H-Index
8
About
Mannes Poel is a versatile computer scientist whose research spans two compelling domains: social touch recognition in human-robot interaction and deep reinforcement learning for autonomous mobile robotics. Perhaps his most influential contribution is the creation of CoST (Corpus of Social Touch), a landmark dataset introduced in 2014 containing nearly 8,000 captures of 14 distinct social touch gestures. This foundational work, which has garnered close to 100 citations across its core publications, established the groundwork for enabling robots and virtual agents to automatically detect and interpret human touch — a critical step toward naturalistic human-robot interaction. His 2015 Touch Challenge further amplified the field's momentum by engaging the broader multimodal interaction research community. Earlier work on gaze behavior with the iCat robot demonstrated his longstanding interest in making robots more believable and socially fluent. More recently, Poel has pivoted toward autonomous navigation, producing a series of papers on deep reinforcement learning and SLAM-based path planning that address how robots can intelligently map and traverse unknown environments. Collectively, his work reflects a consistent commitment to building robots that perceive, interact, and navigate in ways that feel genuinely responsive to their human partners.
Research Focus
Key Achievements
Top Papers
- 1Touching the Void -- Introducing CoST50 citations · 2014
- 2Automatic recognition of touch gestures in the corpus of social touch47 citations · 2016
- 3Touch Challenge '1530 citations · 2015
- 4Gaze behaviour, believability, likability and the iCat25 citations · 2009
- 5
- 6
- 7REINFORCEMENT LEARNING HELPS SLAM: LEARNING TO BUILD MAPS14 citations · 2020
- 8
- 9
- 10