Jasper Hoffmann
PhD Student
Contact
hoffmaja@informatik.uni-freiburg.de
Institut für Informatik
Albert-Ludwigs-Universität
Freiburg Georges-Koehler-Allee 201
79110 Freiburg im Breisgau
Interests
- Rare Events
- Distributional Reinforcement Learning
- Model Predictive Control
- Autonomous Driving
- Active Learning
Teaching
- TA MPCRL Seasonal School, Summer 23
- TA Seminar & Proseminar, Summer 23
- TA Reinforcement Learning, Winter 22/23
- TA MPCRL Seasonal School, Summer 22
- TA Q-learning Proseminar, Summer 22
- TA Reinforcement Learning, Winter 21/22
- TA MPCRL Seasonal School, Summer 21
Publications
2024
Felix Schulz*, Jasper Hoffmann*, Yuan Zhang and Joschka Boedecker. Learning When to Trust the Expert for Guided Exploration in RL. ICML Workshop on Foundations of RL and Control (FoRLaC). 2024. arxiv
Rudolf Reiter, Andrea Ghezzi, Katrin Baumgärtner, Jasper Hoffmann, Robert D. McAllister and Moritz Diehl. AC4MPC: Actor-Critic Reinforcement Learning for Nonlinear Model Predictive Control. 2024. arxiv
Yuan Zhang, Jasper Hoffmann and Joschka Boedecker. UDUC: An Uncertainty-driven Approach for Learning-based Robust Control. 27th European Conference on Artificial Intelligence (ECAI). 2024. arxiv
Jasper Hoffmann, Diego Fernandez, Julien Brosseit, Julian Bernhard, Klemens Esterle, Moritz Werling, Michael Karg, and Joschka Boedecker. PlanNetX: Learning an Efficient Neural Network Planner from MPC for Longitudinal Control. 6th Annual Learning for Dynamics & Control Conference (L4DC). 2024. arxiv
2023
Andrea Ghezzi*, Jasper Hoffman*, Jonathan Frey, Joschka Boedecker, Moritz Diehl. Imitation Learning from Nonlinear MPC via the Exact Q-Loss and its Gauss-Newton Approximation. Proceedings of the IEEE Conference on Decision and Control (CDC). 2023. arxiv
Rudolf Reiter, Jasper Hoffmann, Joschka Boedecker and Moritz Diehl. A Hierarchical Approach for Strategic Motion Planning in Autonomous Racing. European Control Conference (ECC). 2023. arxiv
2021
Jasper Hoffmann*, Shashank Agnihotri*, Tonmoy Saikia, Thomas Brox. Towards improving robustness of compressed CNNs. ICML Workshop on Uncertainty and Robustness in Deep Learning (UDL). 2021. web