Smart Technology : Tracks and traces how avocados move through packhouses

Six smAvo devices were placed in 24 packhouses across Limpopo, Mpumalanga and KwaZulu-Natal during the 2021 harvesting season.

Research results

Processing the more than 12 million data points was a mammoth task that prompted the research team to develop new software to get the job done. Their efforts resulted in the packhouse processing quadrant (PPQ), a system that classifies packhouses according to a damage index score (DIS). The DIS ranges from one (an optimised packhouse) to four (an unoptimised packhouse).

The team furthermore developed a simplified neural network to predict a packhouse’s classification based on nine statistics that are recorded by the smAvo devices. The calibrated model that resulted from the neural network can predict the DIS of a given packhouse with 82% accuracy, outperforming the best linear model.

This calibrated model provides the starting point for accurate, integrated packhouse instrumentation, while simplifying the analytical processes associated with large datasets.

The DIS of a packhouse is an indication of where a facility falls on the spectrum of optimisation, compared to other packhouses.

Research results

Mechanical rollers were one of the most detrimental components in the packline, causing the most severe and longest lasting impacts. The time that the smAvo devices spent being “roughed up” by mechanical rollers varied substantially within a given packhouse, mainly due to the uneven volume of avocados moving through the packline. In general, packlines with slower, smaller rollers delivered a lower DIS score.

According to the DIS, bins are slightly kinder to avocados than crates, although this is likely a by-product of less sophisticated packhouses that process smaller volumes of avocados.

Not surprisingly, staff training emerged as being critical for the better treatment and handling of avocados in packhouses.

The DIS of a packhouse is not inherently good or bad; it is, instead, an indication of where a facility falls on the spectrum of optimisation compared to other packhouses.

What the future holds

The next generation of smAvo instrumentation should be able to classify packhouse performance in real time and relay the information to a cloud-based digital twin of the facility.

This digital twin can be designed to adjust the number of avocados in the packline, and change directed chemical sprays and the speed of the packline to reduce the postharvest stress. With digital twinning, packhouse operations can be adjusted immediately and even in anticipation of the avocados that are about to arrive. It is a vast improvement on the reactive interventions that are currently the only option.

Low-power, long range (LoRa) communications technologies, combined with edge intelligence, are becoming more and more common in agricultural Internet-of-Things (IoT) networks. The smaller these technologies become, the more innovatively they can be used in the fresh produce sector.

Research is currently underway to deploy such technologies to monitor road and sea transportation routes to comprehensively map the farm-to-fork journey of fresh produce. These data-driven investigations promise not only insights for producers, but will also guide road authorities in their quest to improve infrastructure quality to the benefit of the larger community.

Photography by ANDRÉ BROEKMAN