Topic: Safe reinforcement learning

Combine control theory and machine learning for autonomous underwater robotic manipulation at NTNU’s Department of Engineering Cybernetics and Norwegian Centre for Embodied AI. Directions include safe reinforcement learning, physics-informed ML, learning-based model predictive control and foundation models with performance guarantees. Work under Kristin Y. Pettersen and Jan Tommy Gravdahl. The doctorate lasts three years, with a possible teaching-related extension; daily presence in Trondheim is required. Gross annual salary normally starts at NOK 580,000, with a 2% pension contribution. Submi

Navid Azizan presents methods that make neural models obey physical, safety and operational constraints at deployment. Hard-constrained neural networks, or HardNets, enforce input-dependent constraints by construction while preserving universal approximation within the feasible function class. Applications include models of chaotic dynamics with bounded trajectories, energy-constrained operator learning, safe reinforcement learning and control with formal guarantees. The talk then turns to enforcing constraints during sampling from pretrained diffusion and flow-matching models. Formulating gen