Papers

5

Total Citations

51

H-Index

3

About

Johannes Pitz is a robotics researcher pushing the boundaries of **dextrous in-hand manipulation** through **reinforcement learning** and **purely tactile sensing**. His work tackles one of the hardest problems in manipulation: enabling multi-fingered robotic hands to continuously reorient objects using only touch, without external vision or motion capture. Pitz’s major contributions include developing a **modular reinforcement learning architecture** that achieves robust, goal-conditioned reorientation even with the hand oriented upside down—a scenario demanding permanent force closure. His 2023 paper on this topic has garnered **30 citations**, establishing a foundation for tactile-only control. He further advanced the field by identifying and solving the instability issues that arise when naively combining learned controllers with state estimators, detailed in his 2021 work (11 citations). Notably, Pitz demonstrated that a torque-controlled humanoid hand could learn to rotate a cube from scratch using only tactile feedback, a feat previously considered extremely difficult. His recent work (2025) bridges the gap between grasping and in-hand manipulation by using a reinforcement learning critic to compose these skills seamlessly. Pitz’s research is critical for real-world applications where vision is occluded, making his contributions essential for the future of autonomous, dexterous robotics.

Research Focus

Key Achievements

3
H-Index
5
Papers
51
Total Citations
10
Avg Citations/Paper
🏆 Most Cited Paper
Dextrous Tactile In-Hand Manipulation Using a Modular Reinforcement Learning Architecture
30 citations · 2023
📈 Most Prolific Year: 2023 (3 Papers)
🤝 Key Collaborators: 4
🏛 Institutions: Deggendorf Institute of Technology, Deutsches Zentrum für Luft- und Raumfahrt e. V. (DLR)

Top Papers

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5

Key Collaborators

Contact & Links

Available for collaboration
Content generated · 13 days ago