Max Robotics — Global Service Robot Directory + US Certification

About

Thorsten Joachims is a prominent researcher whose work bridges machine learning, robotics perception, and human-robot interaction. He is perhaps best known for his foundational contributions to semantic understanding of 3D environments, particularly the labeling of point clouds captured by RGB-D cameras in indoor scenes. His 2011 paper on semantic labeling of 3D point clouds garnered over 330 citations, establishing key graphical model approaches that enabled robots to meaningfully interpret complex spatial environments such as offices and living spaces. Building on this perceptual foundation, Joachims extended his work to contextually guided object detection and search within 3D scenes, with related publications accumulating nearly 250 additional citations. His research demonstrates a sophisticated understanding of how contextual relationships between objects can guide robotic perception. A second major thread in his work addresses how robots learn human preferences for manipulation tasks. Through co-active and coactive online learning frameworks, Joachims developed methods allowing robots to iteratively refine trajectory planning based on user feedback — a critical capability for personal and industrial robots. His broader vision, articulated in "Learning from User Interactions," advocates for harnessing implicit human feedback across domains from search engines to smart homes. Collectively, his work has shaped both robotic perception and preference learning in meaningful, lasting ways.