For the past five years, if you wanted an automated forklift in your warehouse, it required bolting on a spinning LiDAR dome-a bulky, expensive sensor that shoots lasers to map the facility. LiDAR works great, but it has a massive blind spot: it only sees in 2D planes. If a pallet is sticking out a few inches higher or lower than the laser line, the LiDAR doesn't see it, and the automated forklift will drive right into it.
The industry is now aggressively pivoting to pure vision-guided systems, using standard digital cameras and artificial intelligence. Instead of lasers, these new forklifts have two or three standard cameras mounted on the mast and roof, much like a Tesla. The AI software doesn't try to build a geometric map of the warehouse; instead, it is trained to recognize objects. It looks at a video feed and says, "That is a rack leg. That is a pallet. That is a human."
This changes the economics of warehouse automation drastically. LiDAR sensors cost thousands of dollars and are fragile. Cameras cost almost nothing. Furthermore, if a warehouse changes its layout-moves a rack, adds a new conveyor belt-a LiDAR system often requires a technician to come out and remap the entire facility, a process that takes days. A camera-based AI system simply looks at the new layout, recognizes the new obstacles, and adjusts its path automatically. It handles dynamic environments-like a busy dock where human workers are walking around-infinitely better than lasers. The LiDAR boom in forklifts is peaking, and pure vision is positioned to make it obsolete.
