ATTI asks a selection of industry professionals about their company’s typical testing approach and how it is applied under immense regulatory and engineering pressure, on time and within budget. Next up: Ben Upcroft, VP of technology at Oxbotica.
As an associate professor in robotics with an interest in developing vision systems, Oxbotica’s VP of technology, Ben Upcroft, is responsible for designing the company’s autonomous processes. He says, “The design of autonomous systems must include multiple independent and redundant pathways at the hardware level (sensors, compute and base vehicle), at the software level and at the system level. From software to hardware, automated testing (such as unit, regression, integration, SIL, simulation, HIL, safety, system and structured vehicle testing) forms the basis of our systems analysis before we operate in public domains.”
System verification and validation include the vehicle and accompanying autonomy software, as well as the environment in which the vehicle will operate. The combination of these is known as the operational design domain (ODD).
“Conventionally, verification and validation of an ODD require a vehicle to drive many thousands or even millions of miles,” Upcroft notes. “We’ve introduced virtual verification and validation so that we can be sure the system operates safely in all environments before the vehicle is deployed. Once it enters a particular domain, we receive continuous online feedback that allows us to perform constant verification and validation on the behavior and status of the vehicle’s components, as well as the health of the entire system. This helps us improve our Universal Autonomy Platform continuously.”