Autonomous ice navigation
The marine industry has yet to fully leverage the latest advances in AI and robotics. We're changing that — starting with one of the hardest problems at sea: navigating ships safely and efficiently through ice.
We're building the navigation stack for autonomous Arctic shipping —
reach out to discuss early access or partnerships.
Arctic shipping routes are growing fast — cutting voyage distances by up to 40% compared to traditional passages. But ice navigation remains one of the most demanding challenges in maritime operations, relying heavily on human expertise under extreme conditions.
Advances in AI-driven path planning, real-time sensor fusion, and autonomous decision-making can make ice navigation safer, more efficient, and accessible to a broader range of vessels — while reducing fuel consumption and crew risk.
The team
Second-year PhD student at Mila, in Prof. Liam Paull's Robotics and Embodied AI Lab, where he works on planning and control for autonomous mobile robots. Lead author of AUTO-IceNav, with prior graduate work at the University of Waterloo focused on marine vessel navigation in ice. Recent work includes the top-scoring entry in the 2025 Waymo Vision-based End-to-End Driving Challenge.
First-year PhD student in Computer Science at Mila, working on world models and trajectory optimization for the safe deployment of AI systems in the real world. Background in physics (M.Sc. Astrophysics, B.Sc. Computer Science & Physics) and prior research at Environment and Climate Change Canada on LSTM-based hydrological forecasting.
Associate Professor at Mila and Université de Montréal, and CIFAR AI Chair. His research advances reinforcement learning and optimal control, with a particular focus on the curse of the horizon and how representation learning can overcome it. Applications span energy systems, materials discovery, and drug development.
PhD, PEng. Autonomous systems researcher at Memorial University in St. John's, working on real-time navigation, control, and machine-vision situational awareness for surface vessels in ice-covered waters. Co-author on AUTO-IceNav with hands-on experience running physical trials at the NRC ice tank.
Our team of robotics and AI researchers at Mila – Quebec AI Institute has published peer-reviewed work on autonomous ice navigation, including AUTO-IceNav — a real-time path planning system for ships in ice, validated in physical trials at the NRC Offshore Engineering Basin and published in IEEE Transactions on Robotics (2025) and IEEE International Conference on Robotics and Automation (2023).