Advancing intelligent robotics: SCIoI at the inaugural German Robotics Conference

In addition, SCIoI members Guillermo Gallego and Friedhelm Hamann, along with their colleagues Shuang Guo and Suman Ghosh, will present their latest research at the Conference. Their contributions include two papers on Simultaneous Localization and Mapping (SLAM) using event cameras by Shuang Guo and Suman Ghosh: ES-PTAM, an event-based stereo parallel tracking and mapping system, and Cmax-SLAM, a novel event-based rotational-motion bundle adjustment and SLAM system leveraging Contrast Maximization.

They will also deliver a presentation on their newly accepted CVPR 2025 paper „Event-based Tracking of Any Point“, which builds on their collaboration with the GRASP Laboratory of the University of Pennsylvania, where Friedhelm Hamann recently completed a six-month research stay cofunded by the NeuroPAC scholarship. These presentations highlight the advancements of event-based vision in real-time perception, mapping, and motion estimation for robotics and computer vision applications.

SCIoI member Marc Toussaint and his group members Shiping Ma and Hongyou Zhou will also contribute to the conference with research on optimization, surgical planning, and manipulation strategies. Cornelius Braun and Marc Toussaint will present Stein Variational Evolution Strategies, a novel gradient-free inference method that combines evolution strategies with Stein Variational Gradient Descent (SVGD) to improve optimization in robotics.

Hongyou Zhou will introduce Surgical Planning: A Machine Learning and Optimization Approach, which applies deep learning techniques to reconstruct bone structures from fractured CT scans. Additionally, Marc Toussaint and his collaborators will present Towards Diverse Manipulation Sampling, exploring new methods for generating diverse robotic manipulation strategies without relying on human demonstrations. Together, these works advance optimization, perception, and planning in robotics.

The inaugural German Robotics Conference provides an opportunity for researchers, industry professionals, and policymakers to exchange ideas and explore new developments in AI-powered robotics. SCIoI‘s contributions reflect our ongoing efforts to improve robotic perception, learning, and manipulation through interdisciplinary research. By sharing their latest findings, our scientists aim to contribute to the broader discussion on advancing intelligent systems and their applications.