Trust and assurance—from consumers, the public, and governments—will be critical issues for the AI and autonomous technology space in the coming year. However, gaining that trust will require fundamental changes in the way autonomous systems are tested and evaluated, according to Shawn Kimmel, EY-Parthenon Quantitative Strategies and Solutions executive director at Ernst & Young LLP. . Fortunately, the industry now has access to new methods and innovative methods that promise to revolutionize the field.
The new autonomous environment
Automation has historically been positioned as a replacement for “dull, dirty, and dangerous” jobs, and that continues to be the case, whether it’s work in underground mines, offshore infrastructure maintenance or , prompted by the pandemic, in medical facilities. Removing people from harm in sectors as important and diverse as energy, commodities, and health care remains a worthy goal.
But today’s self-managing technologies are more than just applications, finding ways to improve the efficiency and convenience of everyday spaces and environments, says Kimmel, thanks to innovations in computer vision. , artificial intelligence, robotics, materials, and data. Warehouse robotics have evolved from glorified trams that shuttle materials from A to B to intelligent systems that can freely span space, recognize obstacles, change routes based at the stock level, and can handle dangerous goods. In surgical clinics, robots excel in microsurgical procedures where the slightest human tremor has a negative effect. Startups in the autonomous vehicle sector are developing applications and services in niches such as mapping, data management, and sensors. Robo-taxis are already operating commercially in San Francisco and are expanding from Los Angeles to Chongqing.
As autonomous technology advances into more contexts, from public roads to medical clinics, safety and reliability will become simultaneously more important to prove and more difficult to ensure. Self-driving cars and unmanned air systems have been linked to crashes and casualties. “Mixed” environments, with human and autonomous agents, are known to pose new safety challenges.
The expansion of autonomous technology into new domains brings with it an expanding cast of stakeholders, from equipment manufacturers to software startups. This “system of systems” environment complicates testing, safety, and validation rules. Longer supply chains, with more data and connections, introduce or promote cyber security and risk.
As the nature of autonomous systems becomes more complex, and the number of stakeholders grows, safety models with a common framework and terminology and interoperable testing become requirements. “Traditional systems engineering techniques are pushed to their limits when it comes to autonomous systems,” Kimmel said. “There is a need to test a greater set of requirements because autonomous systems perform more complex tasks and safety-critical functions.” This need, in turn, drives the interest in finding efficiencies, to avoid increasing the cost of testing.
That requires innovations like predictive safety performance measures and preparing for unexpected “black swan” events, Kimmel argued, rather than relying on conventional metrics like mean time between failures. It also requires ways to identify the most valuable and effective test cases. The industry needs to improve the quality of its testing methods without making the process more complex, expensive, or inefficient. To achieve this goal, it may be necessary to manage a set of unknowns in the mandate to operate autonomous systems, reducing the test and safety “state space” from semi-nothing. end to a testable set of conditions.
The toolkit for autonomous system safety, testing, and assurance continues to evolve. Digital twins have become an asset in the development of the space of autonomous cars. Virtual and hybrid “in-the-loop” test environments allow system-to-system testing that includes components manufactured by multiple organizations throughout the supply chain, and reduces cost and complexity. to test the real world through digital augmentation.