Numbers Show How Camera+AI Will Overpower Lidar in AVs (www.getfocus.eu)

A company analysis — developed with MIT and publicized by CEO Jard van Ingen — claims vision-based autonomous driving (cameras plus AI) is improving at an extraordinary >420% rate versus LiDAR’s ~290%, using a patent-derived “rate of improvement” metric. That metric combines cycle time (how quickly inventive steps occur) and knowledge flow (how significant those steps are) to estimate exponential progress much like compound interest: the faster‑improving technology will eventually dominate even if it starts behind. The reported 130 percentage‑point gap signals a durable lead for camera+AI in long‑term capability growth. The significance for AI/ML and AV developers is practical and technical. Camera+AI scales with compute and software advances (Moore’s Law, new neural architectures like CNNs/Transformers, and instant over‑the‑air model updates), enabling huge feedback loops from fleet data — a dynamic exemplified by Tesla’s vision‑first strategy. LiDAR’s advantages (direct 3D geometry, time‑of‑flight reliability) remain valuable, but its progress is limited by physical manufacturing and cost curves. The likely outcome: camera+AI becomes the mass‑market perception backbone, while increasingly cheaper LiDAR pivots to a supporting redundancy role in premium or safety‑critical fleets.