Alphabet‑owned Waymo revealed it has designed a custom computer chip that crunches raw data from cameras, lidar and radar to enable millisecond decision‑making in its driverless cars.
Alphabet’s autonomous‑vehicle unit Waymo announced it has engineered a proprietary computer chip that processes raw inputs from cameras, lidar and radar, delivering millisecond‑scale decisions for its expanding robotaxi fleet.
Why a Custom Chip Matters
Off‑the‑shelf processors struggle to meet the latency and power‑efficiency demands of fully driverless operation. By designing silicon in‑house, Waymo can tailor the architecture to prioritize sensor fusion, reduce bottlenecks, and extend vehicle range on a single charge.
Technical Highlights
The new chip integrates dedicated accelerators for image‑recognition, point‑cloud processing, and radar signal interpretation, all within a single package. Its on‑chip memory hierarchy minimizes data movement, cutting decision latency to under 10 ms.
- Supports simultaneous processing of up to 12 high‑resolution cameras
- Optimized lidar point‑cloud handling for 3‑D perception
- Radar data streamlining for robust object detection in adverse weather
Impact on Waymo’s Fleet
Waymo plans to retrofit its existing robotaxis with the chip while installing it in all new vehicles. The upgrade promises smoother navigation in dense urban environments and more reliable performance in low‑visibility conditions.
Industry analysts expect the move to give Waymo a competitive edge over rivals still dependent on third‑party hardware, potentially accelerating the rollout of fully autonomous ride‑hailing services.