In this lab, students gain hands-on experience with parallel computing principles and the CUDA framework. The tasks include implementing efficient GPU-accelerated computations for robotics and computer vision applications. Specifically, students will first implement a CUDA-based kernel for calculating the Chamfer Distance between point clouds, significantly accelerating this traditionally computationally expensive task compared to its CPU-based counterpart. Additionally, students will explore GPU parallelization of a model-based inspection pipeline, implementing CUDA kernels to accelerate ray casting operations crucial for tasks such as environment mapping and object scanning. More details on the topics and the requirements will be presented at the first introductory meeting.

Prerequisites: C++, Linux