Reproduce “SDPAD: A Fully Spike-Driven Pipeline for End-to-End Autonomous Driving” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Chengjun Zhang, Yuhao Zhang, Jie Yang, Mohamad Sawan
SDPAD makes camera perception and trajectory planning operate through integer-valued spikes in a single feed-forward inference pass. It converts a pre-trained network, uses discrete depth signals to build a bird’s-eye view, and applies learned queries to gather scene context and refine trajectories. On NAVSIM navtest, it scores 86.3 PDMS versus DiffusionDrive’s 88.1, with estimated energy of 34.38 versus 153.22 mJ, not measured deployment.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.