At the recent National Retail Federation’s ‘Big Show’ in New York, the latest technologies and emerging trends supporting traditional bricks and mortar retail were on show. With most attention being paid to e-commerce and with conventional retail having to adapt, evidence of ‘omni-channel’ efforts to fuse aspects of retail and e-commerce were evident.
It’s always a challenge to summarise a major trade event but three major trends shone through:
- consumer tracking and data collection
- inventory tracking and management
- and point-of-sale convenience.
All driven by deep learning, machine vision and artificial intelligence.
Consumer tracking and data collection
The proposition is that the more a store knows about its customers, individually and collectively, the better able it will be to drive sales through personalised messaging and engagement. Vision systems can now track foot traffic and algorithms can help determine age, size, gender, family status and other metrics. Beyond this, start-up companies claim to be able to gauge your:
- interest in an item - based on dwell time;
- purchase intent - based on facial expression; and
- individual style interests - based on what you are wearing.
This data informs retailers' decisions about displays, inventory and promotion. If shoppers ‘opt in’ and allow for the sharing of more personal information via their mobile devices, the shopping experience is further personalized, whilst in-store sales people can be directed to selectively engage individual shoppers with targeted suggestions and promotional offers.
Checkpoint Systems demonstrated its retail system for this situation. Algorithms can also red-flag potential shoplifters who are now deterred by the sudden appearance of an approaching sales person.Having developed machine vision systems for challenging applications, often using sophisticated algorithms, we can leverage inexpensive cameras and processors.
Inventory tracking and management
Having just enough inventory, at the right time and location, is an operational objective to minimise cost and maximise revenue. This may sound easy, but it's a constant challenge for both large big-box retailers like Walmart (which has thousands of SKUs from thousands of suppliers) to single-brand outlet stores like Michael Kors. Attempting to match supply to demand, by taking physical inventory, is typically inefficient, inaccurate and slow - it can take 5-7 days to identify and then replenish an out-of-stock item at Walmart, causing lost sales and customer inconvenience.
Retailers need real-time and accurate inventory control and predictive analytics on rates of adoption for new products. In response to this need, robotics companies like Bossa Nova Robotics are developing automated inventory robots which roam the aisles at Walmart, scanning shelves and reporting inventories. Bossa Nova’s robots can notify management about out-of-stock items within hours rather than days and software can track consumption at the point-of-sale to better inform replenishment requirements.
Similarly, Trax Image Recognition uses stationary on-shelf cameras and shelves with load cells to monitor out-of-stock issues as well as on-shelf quantities. Each of these systems rely on clever vision systems, sensor fusion and algorithms to reliably recognise thousands of SKUs. All of this technology, of course, requires tight integration with back-end inventory software systems but credible real-time inventory tracking is a major step in the progression to just-in-time and better shopper experiences.
Add-up the collective time spent queuing for a cashier and then weighing and/or scanning everything in your cart and you begin to understand why improving the checkout experience is a major goal for retailers. The Amazon ‘Go’ initiative is an indication of where this improved experience lies - on your smartphone and in clever sensing, machine visions and AI. 'Go' informs a near future where you are authenticated by your mobile device, have freedom to browse, evaluate and select items, and then walk out the store with your purchases. In-store sensing and artificial intelligence accurately track your selected items and debit your smart device.
A number of start-ups, such as Bodega, ShelfX and Aipoly, are applying this principle to smart kiosks. There are still several technical challenges to overcome, such as needing to ‘teach’ SKU’s to train the AI, simplifying operator re-stocking by eliminating planogram requirements and developing accurate, robust and fraud-resistant sensing methods but the promise of enhanced consumer convenience and shifting shopping habits are too compelling to ignore.
A new retail reality
Headlines declare that bricks and mortar retail is back on its heels due to the advantages of e-commerce. Improving operational efficiency and the consumer experience are key objectives for retail as it adapts to a new reality. Recent examples of our work with sensing systems, algorithms and AI include Vincent, MyAccent and "Enhancing human capabilities through AR" demonstrations, showing how to engage consumers and provide a richer experience.