Conference Proceeding Talk at IEEE COMPSAC 2021
Date:
In this work, we introduce a GPU-parallel implementation of the Iterative Dynamic Programming (IDP) algorithm for solving the multi-point Markov–Dubins problem, which seeks the shortest bounded-curvature path through several waypoints. Unlike traditional optimization methods (NLP/MINLP), this approach is inherently parallelizable and significantly improves accuracy, speed, and energy efficiency, making it well suited for embedded and real-time applications. Follow the title link for more info.