Papers
12
Total Citations
840
H-Index
8
About
Benedikt Mersch is a leading researcher in robotics perception, specializing in 3D LiDAR-based odometry, mapping, and dynamic scene understanding. His most impactful contribution is **KISS-ICP**, a deceptively simple yet remarkably robust point-to-point ICP registration framework that challenges the trend toward algorithmic complexity in sensor odometry. With over 435 citations, this work has become a gold standard for accurate and reliable ego-motion estimation. Mersch also pioneered learning-based **moving object segmentation** in 3D LiDAR data (237 citations), enabling robots to distinguish static from dynamic elements in real time—a critical capability for safe navigation and consistent map building. He extended this work into **ERASOR2** for instance-aware static mapping and **Radar Instance Transformer** for moving instance segmentation under adverse weather, demonstrating cross-sensor expertise. His recent contributions include **Kinematic-ICP**, which incorporates wheeled robot constraints for planar motion, and self-supervised geometric scan completion for dense mapping. Mersch’s work consistently emphasizes practical, deployable solutions that balance accuracy with computational efficiency, making him a key figure in advancing robust autonomy for mobile robots operating in dynamic, real-world environments.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4
- 5Effectively Detecting Loop Closures using Point Cloud Density Maps27 citations · 2024
- 6
- 7
- 8
- 9
- 10