Immersive Inspection and Annotation of 3D Point Cloud Data in Virtual Reality

Immersive Inspection and Annotation of 3D Point Cloud Data in
Virtual Reality

Background

Modern perception systems generate increasingly rich 3D representations of their surroundings, including LiDAR point clouds, reconstructed environments and semantically labelled 3D data. Inspecting and annotating such data is important for developing, training and evaluating
perception algorithms, but conventional desktop-based tools require users to interact with inherently three-dimensional information through a two-dimensional interface.

Virtual reality provides a fundamentally different way of interacting with 3D data by allowing users to move within the scene, inspect structures from arbitrary viewpoints and interact with objects and regions directly in three-dimensional space. This creates opportunities for new approaches to both inspection and annotation of complex perception data.

The Project

The objective of this thesis is to investigate how virtual reality can be used for efficient and intuitive inspection and annotation of 3D perception data.

The project will explore interaction concepts that take advantage of the immersive 3D environment, including navigation through large point clouds, direct selection and painting of 3D regions, volumetric selection, object-level annotation, and correction of existing semantic labels.

The project will also investigate computer-assisted annotation, where geometric or machinelearning methods propose regions or labels based on limited user input and the user interactively refines the result in VR.
The resulting prototype will be evaluated on representative 3D datasets to assess different interaction concepts in terms of annotation efficiency, accuracy and usability, and to identify where immersive interaction provides benefits compared with conventional desktop-based approaches.

Who We Are Looking For

We are looking for a Master's student in Robotics, Computer Science, Machine Learning, Mechatronics, Engineering, or a related field. You should have an interest in robotics, autonomous systems and perception and enjoy combining research-oriented work with practical implementation.
Experience with one or more areas such as C++, Python, ROS, computer vision, machine learning, point clouds or sensor data processing is advantageous.

What We Offer

At Retenua, you will work at the intersection of research and industrial application, on technology used in real-world robotic and autonomous systems. Our projects are closely connected to current developments in perception, sensing and mobile robotics, giving you the opportunity to work on problems where the solutions are not already predetermined.
As a small and technically focused team, we offer a high degree of independence and the opportunity to take real ownership of your work. You will work with relevant software, sensor data and hardware, and develop and evaluate your ideas in a practical engineering context.

About Retenua

Retenua is a Swedish technology company specializing in advanced perception and mobile robotics. We develop software components for autonomous machines and robots and work with industrial customers on challenging problems involving sensing, perception, localization,
navigation and machine intelligence.

Practical Information

Scope: Master’s thesis, 30 hp
Period: Spring 2027
Location: Örebro/hybrid. A substantial part of the project can be carried out remotely, while meetings, experiments and work involving hardware may require occasional presence in Örebro.
Students: 1-2 students
Application: Please submit your CV, transcript of records and a short description of your interest in the project.