Using AI, IoT to deliver fresh food, cut wastage

According to United Nation’s FAO (Food and Agriculture Organisation) roughly one third of the food produced in the world for human consumption every year (~1.3 billion tons) gets lost or wasted

Updated: Oct 31, 2019 11:13:01 AM UTC

Kedar Kulkarni and Vikas Raykar are Senior Researchers at IBM India.

Image: Shutterstock

Almost every known large industry today produces and distributes goods at scale over a supply chain. In a typical supply chain, as a product travels across the network of supply chain nodes–from the producer, to the intermediate nodes (e.g pre-processor, distributor etc.), all the way to the end consumer–each node adds ‘value’ to the product until it is ready to be sold off at the final node (e.g. retail stores).

Supply chains are customarily designed and operated to minimise costs or maximise profits (or both). One of the most important considerations is to decide ‘when’ and ‘how’ much to replenish each product at each node of the supply chain, as it proceeds from production to consumption. While this customary approach works well for products that ‘do not perish’ or have an unlimited shelf life, a straightforward extension of this approach for perishable products (e.g. fresh produce) will likely lead to significant food waste.

According to United Nation’s FAO (Food and Agriculture Organisation) roughly one-third of the food produced in the world for human consumption every year (approximately 1.3 billion tons) gets lost or wasted. A big percentage of perishables is wasted because it degrades in quality and becomes unacceptably spoilt before consumption at retail stores. At the same time, the demand for perishable food produce is on the rise. Today’s consumers are better informed and are influenced by marketing campaigns that recommend nutritious and fresh food products. Data indicates that consumers are fast switching to fresher options and shying away from pre-processed packaged food.

Yet, most retailers are grappling with the challenge of consistently providing fresh food to the customers while still making a profit. Freshness is clearly on the agenda for major retailers: Walmart’s recent program about a 100 percent money-back on unacceptable quality of fresh fruits and vegetables is a case in point.

In the case of fresh produce, the perishability clock starts ticking the moment the produce is harvested in the farm. The ‘freshness’ of this produce is measured by the so-called ‘Remaining Shelf Life’ (RSL) metric. The RSL is the highest right after harvest and decreases as the produce travels across the supply chain. The higher the post-harvest time-delay, the lower the RSL is at consumption. Another crucial factor that affects freshness is the post-harvest temperature and humidity levels that the food is exposed to. Higher the temperature/humidity, the lower is the RSL. It must be mentioned that, while the degradation rate is, in general, different for different kinds of food, post-harvest time-delay and temperature/humidity conditions are known to be the most crucial factors that affect the freshness of all kinds of perishable food.

Thus, an ideal supply chain strategy would ensure that (a) fresh produce reaches the stores (and ultimately the end consumers) fast (i.e. before the RSL runs out) while ensuring that the food (b) is maintained at the optimal temperature/humidity throughout its supply chain journey.

The success of such a strategy can be largely enhanced by using emerging digital technologies like artificial intelligence, internet of things and blockchain.

AI for inventory planning
Traditional inventory planning does not account for perishability of goods. A perishability-aware inventory planning system would, among other things, effectively optimise the fresh food replenishment order-size and frequency, while accounting for freshness criteria and wastage constraints.

Internet of Things and blockchain for traceability and tracking
Food retailers are increasingly using RFID (Radio Frequency Identification) tags to track food as it travels through the supply chain network. These tags can monitor information such as the stock serial number, expiry dates, and so on. At the same time, important supply chain events can be recorded immutably on a blockchain, that entities on the network can access on a need-to-know basis.

AI for RSL estimation
Currently, all fresh produce harvested at the same time is assumed to have the same RSL, irrespective of the post-harvest temperature/ humidity conditions experienced by the food as they travel over the supply chain. However, temperature/ humidity sensors installed on food packets can potentially provide data to estimate and monitor RSL more accurately for different foods.

AI for dynamic real-time allocation
Using IoT-based tracking information and the estimated RSL, food can be dynamically allocated to different nodes of the supply chain, based on the ‘match’ between RSL and demand. For example, a food with a very low estimated shelf life on its way to a far-off destination would be dynamically rerouted to a closer destination with a high demand, where it will likely be consumed very quickly, thus avoiding food waste.

While the above list of technology enablers is not exhaustive, integrating these into the food supply chain strategy could potentially (i) improve the quality and freshness of the delivered perishable food and will (ii) reduce food waste.

The writers are senior researchers at IBM India.

Post Your Comment
Required, will not be published
All comments are moderated