Argo AI has announced that its testing procedures for autonomous vehicle technologies now conform with the standards set by TÜV SÜD. The latter is a world-leading certification body in the autonomous vehicle industry that recently conducted an audit of operations at Argo AI. This was to verify that Argo AI’s methods in test driver selection, training, and oversight conform to the practices set by the Autonomous Vehicle Safety Consortium (AVSC) and SAE International’s J30188 standards for safe on-road testing.
TÜV SÜD’s evaluations found that Argo AI met and, in some cases, exceeded the industry best practices and standards outlined by AVSC and SAE International’s J30188. It cited Argo AI’s four-week “Autonomous Vehicle System Test Specialist” certification course as compliant with the above applicable standards. “Safety is the foundational value at Argo AI,” said Bryan Salesky, Founder, and CEO of Argo AI. “When testing autonomous vehicles, we want to ensure the safety of not only our employees but also the local residents with whom we share the roads. There are no shortcuts when it comes to developing safety-critical products.”
Test specialists or highly trained operators of autonomous vehicles conduct the development of Argo AI’s self-driving systems. They oversee the driving capabilities of the company’s global fleet of autonomous vehicles, which are now undergoing testing in six cities in the U.S. and two in Germany. Argo AI voluntarily had the audit done by TÜV SÜD to confirm that its practices, methods, and efforts were consistent with the industry standards.
Christian Gnandt, Vice President of the Automated Driving Division at TÜV SÜD, noted that there are currently no statutory requirements for third-party assessment before a company receives permission to test autonomous vehicles on public roads in the U.S. “The fact that Argo still approached TÜV SÜD for assessment is a clear demonstration of its commitment to autonomous vehicle safety, and its trust in our expertise in the autonomous vehicle field,” added Gnandt. “We were very happy to work with Argo and to have the opportunity to draw on our extensive knowledge in this field and also expand our experience at the same time.”
The audit initially looked at Argo AI’s hiring and training process for the Test Specialist role. Candidates must have a valid driver’s license, pass criminal, education, and employment verification background checks, screenings for DMV records, and a 10-panel drug screen and alcohol test. Once those are done, they can start the four-week training course, which includes classroom instruction, test track training, and on-road evaluations. It also includes “fault injection” training, a process where Argo creates unexpected scenarios on the test track so that the drivers can react accordingly if they encounter the same experience in real-world situations.
Once that’s done, the audit continues by looking at how Argo AI prepares a fully trained Test Specialist for daily missions. These start with “pre-mission briefings” or the safety procedures that all Test Specialists must undergo each day before getting behind the wheel. It is then followed by Argo’s “on-mission protocols,” which include a 15-minute break mandate every two hours of driving to stay focused and minimize fatigue. TÜV SÜD also noted that Argo AI uses a driver monitoring system via a dashboard camera that tracks the driver’s attention level and ensures they’re staying focused on the road ahead.
Lastly, Argo AI also detailed its “post-mission protocols,” including how Test Specialists give feedback on vehicle performance via shift summaries. This is also the time where the driver logs and escalates system issues to operations managers who will send the info to the engineering team. Test Specialists undergo a continuous performance review to meet ongoing operational standards. Regular training sessions teach new skills while an annual multiday recertification program ensures that they have the knowledge and compliance. Salesky notes that public trust is just as important as delivering autonomous technology. “We can develop the technology, but if consumers and communities don’t trust it, then we’ve not done our job,” he noted. “We hope this extra step helps ensure that Argo AI is earning the trust needed to have self-driving cars welcomed to the market.”