Jan Gutsche

Papers

Total Citations

H-Index

About

Jan Gutsche is a robotics researcher whose work centers on embedded computer vision and real-time perception for autonomous systems, particularly humanoid robotics. His most impactful contributions lie in developing efficient, hardware-constrained visual processing pipelines. Gutsche’s 2019 paper on an open-source vision pipeline for RoboCup humanoid soccer (7 citations) provides a foundational framework for robots to perceive and react in dynamic environments using limited onboard computing. His landmark 2021 work, “YOEO–You Only Encode Once” (7 citations), introduces a novel hybrid convolutional neural network that unifies object detection and semantic segmentation through a single shared encoder backbone. This architecture dramatically improves inference speed and accuracy on embedded devices, directly addressing the critical challenge of deploying deep learning on resource-constrained robots. By enabling mobile platforms to perform complex visual tasks without sacrificing performance, Gutsche’s research has practical implications for everything from autonomous navigation to human-robot interaction. His work exemplifies the push toward lightweight, real-time AI that makes intelligent robotics more accessible and responsive in the physical world.