Certification-driven autonomy for defense
Scott Sanders, chief growth officer at Forterra, on the defense tech startup playbook
The real bottleneck in autonomous ground vehicles is not making the vehicle move, it is proving to a certifier that the same sensor inputs lead to the same safe action every time. That is why Forterra emphasizes tightly controlled autonomy stacks instead of generative models. Safety approval in autonomy is built around a safety case, meaning documented evidence from analysis, simulation, and testing that the system behaves within defined limits, not around a model that may answer slightly differently from run to run.
-
Forterra built its product around a repeatable kit that can be installed across vehicle types, from military trucks to yard tractors. That common hardware and software base matters because certification gets easier when the company is validating one core system repeatedly, instead of a different stack for every program.
-
The standards direction in autonomy reinforces this view. UL 4600 is centered on a safety case for autonomous products, and its guidance points to evidence from simulation, closed course testing, and road testing to show acceptable safety. That framework fits deterministic, bounded systems better than open ended generative behavior.
-
This is also why off road autonomy is harder than passenger car autonomy. Forterra describes ground defense autonomy as a 3D problem across mud, snow, rain, hills, and unknown terrain. Each extra environment creates more edge cases that have to be shown safe, and the last few percent of reliability carries most of the certification burden.
Going forward, the winners in autonomy will look less like chatbot companies and more like industrial system builders. The advantage will come from owning the sensor stack, vehicle interfaces, test data, and safety evidence needed to clear procurement and regulatory gates. In defense and heavy industry, certifiable consistency will matter more than flashy model demos.