Autonomous Retailing: A Frontier for Cyber-Physical-Human Systems

Retail is one of the largest economic sectors, accounting for almost $5 trillion in sales in the US alone. With the proliferation of e-commerce, mobile devices, and digitally engaging shopping journeys, retail is going through profound transformations that will change everyone’s life. The future of retail will inevitably integrate online and in-store shopping, and promises to enhance customers’ shopping experience. Physical stores, which still account for 85% of retail sales, and 95% of grocery sales, must be repositioned to coexist with online and mobile shopping channels. Autonomous retailing is a retail process where a physical store is aware of all elements involved—products, people, and activities—without explicit help from human workers. Autonomous stores allow shoppers to pick up products and walk out of the store, without going through a checkout lane. Although the concept is more than a decade old, Amazon Go, a recent effort to realize frictionless checkout, brings it a huge step closer to reality. Autonomous stores are an example of cyber-physical-human systems that incorporate advanced artificial intelligence (AI) through abound embedded sensors and computation. Natural human activities bring significant challenges to system provisioning, sensing, and inference, but also provide input for the system to learn from and adapt to. In this article, we discuss the design space and technical challenges of autonomous retailing and motive it as a frontier of cyber-physical-human system research.