[6 months, 4-5 days/week] Name of Project: Pangu Objectives: - Mercedes-Benz are performing the algorithm study and software development based on multi-sensor data (LiDAR, cameras, etc). The intern will work as an 4D reconstruction engineer in our internal data driven auto-labeling pipeline and software development, together with team leader and other algorithm engineers. The intern will be guided by team leader to do start-of-the-art algorithm research with our internal, large scale dataset. Main Tasks: - 4D spatiotemporal scene reconstruction and generation algorithms such as SLAM, NeRF, 3DGS based on multi-sensor data (LiDAR, cameras, etc.) - Necessary data processing on the sensor and dataset data - Necessary tools development for the research Learning Opportunities: - Work on the 4D scene reconstruction/generation research and development with the large scale MB internal dataset - Develop start-of-the-art 4D reconstruction method - Opportunities to work on the project as a pure developer with MB algorithm and software engineering team - Explore and validate new research ideas on the MB AD RD platform Key Qualifications: - Understanding on LiDAR sensor principle. Experience of point cloud processing; CV, multi-view geometry, SLAM, NeRF and 3D Gaussian Splatting. - Familiar with Python programming; PyTorch, experience of building a data preprocess, training and model exporting python packages - Foundation in linear algebra, probability theory, and optimization theory - Research background on joint optimization, multi-traversal Gaussian Splatting; on AI, CV or Robotics - Familiar with C++ and ROS programming, familiar with ROS related tools usage; 3D Gaussian Splatting (3DGS) and Neural Radiance Fields - Development or research experience on CUDA and TensorRT programming - Hands on experience on AD car/system development - Top conference paper is a plus Education: Degree: Master Major: CS, Robotics, Electronics