Puzzle solving autonomy for yards
Scott Sanders, Chief Growth Officer at Forterra, on autonomy for every vehicle
This reveals that Forterra is betting ground autonomy will be won by systems that can reason through one off physical situations, not just replay moves seen in training data. In a yard or railhead, trailers, containers, people, and blocked paths change every day, often with weak connectivity and poor GPS. That makes autonomy look less like consumer self driving and more like defense robotics, where the software has to keep working when the map is wrong and the scene is messy.
-
Forterra chose terminal tractors because the economics are tighter and the deployment problem is simpler. Large yards cluster many vehicles in one place, so a company can supervise early rollouts with fewer people, integrate once with a dominant OEM, and spread the autonomy stack across many vehicles at the same site.
-
The product is not just a driverless truck. Kalmar supplies the factory vehicle and yard automation hooks, while Forterra supplies AutoDrive, the onboard system that senses obstacles, plans paths, and keeps operating in mixed traffic and all weather. That is a concrete example of puzzle solving embedded in an industrial workflow.
-
This same design logic shows up in defense. Forterra describes a distributed system that cannot depend on a central brain or perfect communications, and Anduril is pushing a similar integrated software plus hardware model for robotic combat vehicles. The common requirement is autonomy that keeps functioning when infrastructure is limited and conditions keep changing.
The next step is more autonomy in places that look like yards before it reaches open roads at scale. As compute and sensor costs keep falling, the winning companies are likely to be the ones that turn rugged, factory integrated autonomy into a repeatable product for ports, depots, and military logistics, then expand outward from those controlled but constantly changing environments.