What AI says about GRaD-Nav

GRaD-Nav is a research-oriented project repository that presents methods for learning-based visual drone navigation using Gaussian Radiance Fields and differentiable dynamics. It offers code, data assets, and setup instructions (including Nerfstudio, point cloud data, and 3DGS data) to train and evaluate navigation policies on drones. The project ships with training scripts and references to project pages and arXiv publications GRaD-Nav and GRaD-Nav++ that demonstrate state-of-the-art visual navigation for unmanned aerial vehicles.