OctoSense: Self-Supervised Learning for Multimodal Robot Perception

arXiv·medium signal

Releases an open-source sensor platform combining stereo RGB and event cameras, LiDAR, a thermal camera, an IMU, and RTK-corrected GPS, paired with self-supervised learning for multimodal robot perception. The open hardware-plus-method stack lowers the barrier to building perception systems. Notable for robotics builders wanting a reproducible multimodal sensing baseline.