PhD in 3D Dynamic Scene Understanding and Scene Flow at KTH Royal Institute of Technology KTH Royal Institute of Technology in Sweden

PhD opportunity at KTH Royal Institute of Technology in 3D dynamic scene understanding, with a focus on scene flow estimation, robotics, perception and learning, and autonomous systems.

The project addresses a key challenge in modern perception: making 3D motion understanding generalize across different sensors and environments, while also supporting the safety and decision-making needs of autonomous systems. The research direction is strongly oriented toward self-supervised and multi-modal learning, including methods that combine LiDAR geometry and camera semantics to learn robust motion representations.

The position is hosted by the Department of Robotics, Perception and Learning at KTH and is linked to the Wallenberg AI, Autonomous Systems and Software Program (WASP). The student will also be connected to the WASP graduate school. The starting point for the project is the OpenSceneFlow codebase and related research on self-supervised scene flow.

Funding: fully funded PhD position with monthly salary according to KTH’s doctoral student salary agreement.

Eligibility highlights: a second-cycle degree (for example a master’s degree) or equivalent, or at least 240 higher education credits including 60 second-cycle credits; English equivalent to English B/6; strong ability to program and implement machine learning methods, preferably in computer vision; and good independent, collaborative, and analytical skills.

Application: apply through KTH’s recruitment system. The application must include diplomas and grades, proof of language requirements, CV, a one-page motivation/application letter, and a list of publications or technical reports. Deadline: 2026-05-15.