How Suning’s Fourth‑Gen AI‑Powered Visual Unmanned Stores Transform Retail

Suning describes the next decade as the "scene retail" era and shares its store digitalization capabilities through a series of lectures, focusing on the fourth‑generation fully digital visual unmanned store technology.

The retail industry has undergone three major waves: the chain‑store era, the internet era, and the current digital retail era driven by mobile payments, AI, and cloud computing, which blur the line between online and offline.

Over the past 30 years, Suning has participated in each wave, evolving from a 1990s air‑conditioner specialist to a 2000s appliance chain, launching Suning.com in 2010 for internet transformation, and entering the smart retail era in 2017 with cloud, big data, and AI technologies.

Suning’s digitalization journey is divided into five stages: early information‑technology development (1990‑1999), ERP era with SAP (2000‑2009), the "+Internet" era (2010‑2014) with the launch of Suning.com and global R&D centers, the "Internet+" era (2015‑2016) establishing a three‑layer architecture and AI labs, and the smart retail era (2017‑present) where store digitalization began.

Today Suning employs over 10,000 technical staff across research bases in Nanjing, Beijing, Shanghai, and Silicon Valley, handling massive data workloads (200 billion services, 42.35 PB daily processing, 34.65 trillion calculations) and operating an AI platform with 300+ skills across 80+ retail scenarios.

Suning’s store digitalization addresses four dimensions—customer, product, store, and operations—transforming traditional challenges such as limited acquisition channels, low user value extraction, inefficient supply‑chain coordination, weak product management, and poor store efficiency into opportunities for multi‑channel acquisition, AI‑driven supply‑chain optimization, precise product management, O2O integration, and data‑driven decision making.

The digital store workflow includes intelligent site selection using big‑data and AI, AI‑guided product planning, smart checkout robots, facial‑recognition tagging, interactive tools (smart mirrors, AR, smart carts), and seamless post‑sale analytics for traffic, heat‑maps, and remote monitoring.

Suning’s unmanned store evolution spans four generations: the 2017 Biu store (first unmanned checkout), 2018 Biu 2.0 with AR and crowd analytics, 2019 Biu × 24h dual‑mode store, and the 2019 fully digital visual store that adds comprehensive real‑time detection of customers and products using both low‑cost CPU‑based traditional image algorithms and high‑cost GPU‑based deep‑learning solutions.

The fourth‑generation store combines 3D cameras, high‑precision visual algorithms, and multi‑sensor fusion to achieve full‑scene, full‑time detection of people and goods, enabling AI‑driven marketing, real‑time virtual carts, and millisecond‑level billing as customers leave the store.

The system architecture is divided into hardware (cameras, sensors, servers), perception (3D reconstruction, tracking, product recognition with visual‑weight dual verification), and decision layers (virtual cart management and checkout), delivering a seamless in‑store shopping experience.

Future plans include two scalable solutions: a low‑cost CPU + security camera + traditional algorithm stack for simple stores, and a hybrid CPU/GPU/ARM + 3D camera + deep‑learning stack for complex environments, allowing flexible cost‑performance trade‑offs across Suning’s extensive store network.

By capturing detailed in‑store user trajectories and product interactions, Suning can apply online‑style recommendation algorithms to the offline environment, dramatically improving marketing precision and conversion rates.