David BECK analyzes technological issues from a political, economic and management perspective.

Walmart Deploys AI-Powered Self-Service Scales

David BECK Academic - Society, Politics & Techology

Customers now benefit from significantly faster checkout experiences. As the largest retailer in the world, Walmart is one of the leader in leveraging transformational technologies to improve the customer shopping experience, reduce costs and enhance efficiency. Walmart uses a communication platform for connecting Walmart and leading technology companies. It works as a bridge between technology companies and Walmart to jointly create technical solutions that can solve for the pain points of retail.

The platform has selected only a handful of companies that meet the high standards of Walmart. The best of those projects will be deployed in Walmart stores to be used for optimizing store operations and improving customer experience.

Encouraging open collaboration

In response to increasing customer demand for self-service shopping, Walmart is leading the way by launching self-service fresh produce scales in many stores. This improves the shopping experience by saving customers’ time and making it easier for them to shop, all while lowering operational costs. With conventional self-service scales, customers often needed to click through several screens to identify the correct product among hundreds of possible options. This can be confusing, often leading to customers requiring associate assistance to successfully complete the weighing process.

Designed to decrease the amount of help consumers need from associates during the self-checkout process, the platform can ID products even when they’re placed inside clear plastic bags. Shoppers can then select their items with a single tap vs. the previous method involving multiple screens. Machine learning enables the technology to continuously improve its capabilities by recognizing new SKUs and processing data without employee intervention. Average weigh time has been reduced by about 40%, according to data from 10 Walmart stores, with higher customer satisfaction and increased overall transactions.

A more positive customer shopping experience

The AI-powered self-service scale solution integrates product recognition. The technology seamlessly integrates into Walmart’s already-deployed traditional self-service scales, transforming them into cutting-edge AI devices, minimizing Walmart’s need to purchase new equipment.

During use, customers simply place the bagged items (e.g. fruits, vegetables and other loose/dry goods) on the scale and intuitively, the AI does the rest of the work. The system accurately identifies the product without any change to normal customer or shopping behavior – this even includes items within multiple layers of clear plastic bags or other protective materials.

Customers are greeted with a “one-tap” ready ranked list to select on screen. In the past, there were multiple screens and taps involved, which increased friction in the shopping experience. Now, with just one click, customers are empowered with a far simpler and quicker experience to print their payment label and complete the process.

Apply AI product recognition

Normally, an AI-based product recognition system requires data to be collected, labeled and trained in a separate process involving significant manual and high-cost effort. In this traditional machine learning approach, store associates would need to be involved for data collection, professional annotators called upon for data labeling, and high caliber data scientists for model training. This leads to significant inefficiency because this process requires continuous updating as new products are introduced to the store.

This new approach enables the automatic learning of new SKUs, without requiring any manual human intervention and becomes smarter over time in a positive feedback loop. In order to help people live better through Corporate Social Responsibility (CSR) initiatives, the solution is carefully designed to protect customer privacy. The camera involved points downward, only capturing the produce placed on the scale within its field of view, not the shopper. Additionally, since the system is based on self-learning, all data processing is done automatically by the system itself.

From a software perspective, the solution leveraged a trailblazing “weakly supervised learning” algorithm, which was integral in enabling it to achieve self-learning. It leveraged NVIDIA T4 Tensor Core GPUs for AI inference in combination with the NVIDIA EGX optimized edge software stack running on Dell EMC PowerEdge servers.

In the first wave of deployment, the solution is being launched in 100 Walmart stores in China. This is providing a vast number of Walmart customers an opportunity to experience the benefits that technology innovation can bring to everyday living. Today, the solution has been production-deployed throughout Walmart stores.

Zeen is a next generation WordPress theme. It’s powerful, beautifully designed and comes with everything you need to engage your visitors and increase conversions.