PhD: Safe, Long-Horizon Robot Planning Under Uncertainty

Delft; Delft; On-site; EUR 34,000 - 43,000; Full time

Job summary

TU Delft's Cognitive Robotics Department invites applications for a PhD on Safe Planning and Decision-Making under Uncertainty for Long-Horizon Robot Autonomy. You will develop planning methods that let robots act safely over long time horizons despite incomplete knowledge of the environment.

The work combines model-based planning and learning, analyzes guarantees and failure modes, and is validated in simulation and on real robots, in collaboration with EU partners in OPERA.

Qualifications

• MSc degree in Systems and Control, Robotics, Applied Mathematics, or related field.
• Interest in safe planning and decision-making for robots at the intersection of planning, control, and learning.
• Rigorous, theoretically grounded research with ability to validate ideas in simulation and on real robots.
• Strong analytical and mathematical skills in optimization, probability, statistics, and dynamical systems.
• Excellent communication skills and proficiency in English (written and verbal).
• Experience with optimization-based planners or controllers (e.g., MPC with acados or CasADi) in C++ or Python.
• Experience in safe or robust control and planning under uncertainty.
• Knowledge of learned dynamics or model-based reinforcement learning is a plus.

Responsibilities

• Develop planning and decision-making methods for safe long-horizon robot autonomy.
• Analyze guarantees and failure modes of proposed methods.
• Collaborate with OPERA partners across Europe on simulation, perception, and toolboxes.

Skills

Analytical & mathematical skills

English proficiency

Education

MSc degree in Systems and Control, Robotics, Applied Mathematics, or a closely related field

Job description

TU Delft's Cognitive Robotics Department invites applications for a PhD on Safe Planning and Decision-Making under Uncertainty for Long-Horizon Robot Autonomy. You will develop planning methods that let robots act safely over long time horizons despite incomplete knowledge of the environment.

The work combines model-based planning and learning, analyzes guarantees and failure modes, and is validated in simulation and on real robots, in collaboration with EU partners in OPERA.