Master Thesis: Long-term Global Mapping for Autonomous Outdoor Navigation

Call for applications for the field of study such as: Computer Science, Software Engineering, Mechanical Engineering, Mechatronics or similar. In the research group Navigation Mobile Robots, we develop autonomous, mobile robots for a variety of outdoor applications, such as agriculture, forestry and logistics. We focus on developing both an autonomous outdoor navigation solution and the hardware of the robots. Be part of change Maintaining and updating a consistent global costmap is essential for mobile robots operating in big, unstructured and highly-dynamic outdoor environments, as they allow the robot to plan long-range, environment-aware routes and avoid obstacles effectively over extended distances. The environments where the robot operates are expected to experience large temporal variations due to weather conditions, seasonal changes, new/modified infrastructure, and dynamic obstacles. Therefore, the mapping approach must adapt accordingly, distinguishing between permanent structural changes and temporary variations, ensuring that only relevant static features are integrated over time. The development of a robust mapping framework presents two main challenges. First, it must handle the uncertainty and temporal inconsistency of the localization estimates. Loop closures or other optimization steps in the localization module can introduce drastic corrections to the absolute poses of previous timestamps, requiring the mapping framework to update the map accordingly. Second, it must handle dynamic obstacles and temporary changes, ensuring they are not added to the global static map representation. The following master thesis aims to resolve the previously stated challenges, developing a robust mapping framework that allows for accurate and consistent online map growth during exploratory tasks while efficiently adapting the map to reflect structural changes during long-running robot operations. n this master thesis you will develop a robust mapping framework capable of handling localization uncertainty and pose corrections, dynamic environments and long-term changes. You will evaluate different submap-based representations that allow for incremental map growth during exploration while ensuring consistency with the robot’s estimated trajectory from the localization module. You will also evaluate different map update strategies that enable adaptation to permanent structural changes while remaining robust to temporary or dynamic variations. You will evaluate the approaches based on scalability and memory and computation efficiency for the online use on the robot. You will then implement your solution inside our ROS2, C++ navigation stack and integrate it with existing components such as the LiDAR-Inertial localization module and the Nav2 Framework. Its performance will then be tested both in simulation and in real-world scenarios with our mobile CURT robots.