Researcher (Postdoc) in Off-Road SLAM and Advanced Perception USP – Unique selling point

Are you passionate about autonomous systems and advanced perception techniques? We offer an exciting postdoctoral position, in collaboration with Demcon, focusing on off-road SLAM using lidar, camera, and IMU.

Do you have a deep understanding of semantic segmentation, SLAM, sensor fusion (lidar, camera, IMU), embedded programming, and/or real-time computing? Are you ready to push the boundaries of autonomous navigation in challenging environments like forests and off-road terrains? Join our innovative project where simulated data and real hardware testing create innovative solutions.

The work builds on our recent results.

You will be at the forefront of developing advanced perception systems, using deep learning and simulation-to-real domain adaptation techniques. You will work with a multidisciplinary team, contributing to fundamental and applied research. Your role will involve designing and implementing algorithms. Testing with actual robotic platforms is handled by Demcon. Innovative simulation data is used for both training and testing.

The project is in parallel with our dedication to teaching perception in our MSc Robotics programme and MSc Geo-information programme, offering you also the chance to mentor and inspire the next generation of engineers (all teaching is fully optional). This is a unique opportunity to blend academic research with real-world applications, supported by a leading company in the industry.

After successful employment at the University of Twente, it is possible to discuss a (part-time) position at Demcon.

• A PhD degree in a relevant field.
• A consistent track record of research excellence.
• Knowledge of and experience in Scan matching and odometry in unstructured/off-road environments, Semantic segmentation of LiDAR and camera data using modern AI methods, and Geometry-semantics fusion for robust navigation and mapping (e.g., representation learning, distillation, feature/attention alignment, joint perception-SLAM models).
• Passion for research and algorithms, including Python/ONNX and C, e.g. pytorch and ROS2.
• Excellent communication skills and a strong command of English.

The following are a plus:

• Simulation-to-real domain adaptation and experience on related techniques.
• Research & development experience in C-based deep-learning inference on GPU and CPU (e.g., ONNX Runtime, TensorRT, OpenVINO, or similar).
• Solid understanding of high-performance optimisation and model fine-tuning techniques.
• Strong knowledge of ROS2 architecture, including executors, callback groups, and real-time considerations.

For more information about the position, please contact Dr Ville Lehtola (email: v.v.lehtola@utwente.nl). You are also invited to visit our institute [homepage](https://www.itc.nl), and the [robotics centre homepage](https://www.utwente.nl/en/robotics/).

For questions about working and living in the Netherlands please consult [the official website of the Netherlands Government](https://www.netherlandsworldwide.nl/theme/work-in-the-netherlands) or the website of the [Expat Center East Netherlands](https://expatcentereastnetherlands.nl).

Please submit your application before 16 January 2026. Your application should include:

• A motivation letter (maximum 2 pages of A4), emphasising your specific interest, qualifications and motivation to apply for this position.
• Curriculum Vitae with a full list of publications, and two references with their contact details.
• A short description (150 – 250 words) of your PhD research and a copy of your PhD diploma. If you are in the final stage, then a letter from your PhD supervisor, stating the date of PhD defence, is also sufficient.

Interviews will be held before the 30th of January 2026. If possible, the employment will start on the 1st of March 2026.