Reproduce “Towards Real-Time Full-Waveform LiDAR Transformers via Intensity-Guided Token Reduction and Physics-Aware Augmentation” 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
Kotaro Oishi, Kazuma Ikeda, Ryosei Hara, Ryo Yoshida, Mariko Isogawa, Kentaro Yoshioka
FWL-ToPM accelerates full-waveform LiDAR perception by pruning low-intensity patches and merging remaining representations while retaining their spatial organization. On Ghost-FWL voxel-wise ghost detection, it achieves 0.592 F1-mean at 53.2 full-frame FPS, versus 0.503 at 6.54 FPS for prior FWL-MAE. This is model-level throughput on one GPU, not end-to-end robot latency; pruning can discard weak real returns.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.