Thesis - Dynamic Obstacle Detection and Prediction for Autonomous Outdoor Navigation
Fraunhofer-Gesellschaft · Stuttgart
What you'll do
You will first select a suitable state-of-the-art approach for detection and classification based on LiDAR and/or camera data. Your focus, however, is on tracking and prediction of these obstacles and integrating this information into a costmap. For this, you will investigate various types of prediction methods, such as intent prediction, reachability approaches, or Bayesian filtering, and develop a robust prediction algorithm. These dynamic predictions must then be integrated into a map representation so that they can be considered and used in the robot's path planning. You will implement your solution in our ROS 2/C++ navigation stack and validate it both in simulations and in real-world scenarios with our mobile CURT robots.
What you need
- Valid enrollment at a German university or college
- Programming skills and practical experience with C++/Python
- Experience with ROS is an advantage
- Analytical mindset
- Enthusiasm for mobile robotics
- Regular presence at our location directly on the campus of the University of Stuttgart
- Fluent German or English skills
What you get
- Cutting-edge technology in the field of outdoor robotics
- Practical work with our robots in Stuttgart
- Personal responsibility and freedom to implement your own ideas
- Collaboration with the best students in their field