Edgar Welte
Papers
2
Total Citations
5
H-Index
2
About
Edgar Welte is a leading researcher in dexterous robotic manipulation, with a focus on developing robust, uncertainty-aware learning methods for real-world applications. His work bridges the gap between imitation learning and probabilistic grasp planning, addressing critical challenges in humanoid robotics. Welte’s major contributions include pioneering evidential learning approaches for contact-grasping in noisy, cluttered environments, as exemplified by his highly cited paper “vMF-Contact: Uncertainty-Aware Evidential Learning for Probabilistic Contact-Grasp in Noisy Clutter” (2025, 3 citations). This work uniquely models both aleatoric and epistemic uncertainty, enabling robots to safely handle out-of-distribution objects and sensor noise—a key advancement for deployment in unstructured human-centric spaces. Additionally, his comprehensive survey “Interactive imitation learning for dexterous robotic manipulation: challenges and perspectives” (2025, 2 citations) synthesizes the field’s state-of-the-art, outlining future directions for sample-efficient, adaptable manipulation. Welte’s research has immediate impact on humanoid robotics, where precise and safe interaction with everyday objects is paramount. His work is essential reading for students and researchers aiming to build robots that learn from human demonstration and operate reliably under real-world uncertainty.
Research Focus
Key Achievements
Top Papers
- 1
- 2