JoachimClemens/Evidential-FastSLAM

Evidential-FastSLAM is an academic-grade project that is over 10 years old. While it explores a technically interesting niche—using belief functions (Dempster-Shafer theory) instead of standard Bayesian probability to handle ignorance in map building—the project lacks modern relevance. With only 22 stars and zero recent activity, it serves as a historical reference implementation rather than a viable production tool. The SLAM field has moved significantly toward Visual SLAM, Lidar-based SLAM (like Google Cartographer), and more recently, neural-network-based mapping techniques like NeRFs and 3D Gaussian Splatting. Frontier labs are unlikely to compete here because the industry has effectively moved past particle-filter-based occupancy grids for high-performance robotics. Defensibility is minimal; the value lies solely in the specific mathematical implementation of evidential reasoning for FastSLAM, which is easily reproducible by researchers in the field.