Developing flexible automation for mushroom harvesting (Agaricus bisporus) : innovation report

Abstract

A framework for analysing crop processes and their suitability to automation was
\ndeveloped in order to address the challenges of labour costs and skills availability that
\nUK growers face. Harvesting was found to be the function of greatest potential labour
\nresource savings. The framework compared those crops with the highest Home
\nProduction Marketed value, in terms of target detection, target removal, seasonality and
\nenvironmental factors. Agaricus bisporus (common mushroom) was the crop that was
\nidentified as the best candidate for automation.
\nTherefore a laboratory demonstration of a robot arm was designed and developed and
\nexperiments conducted showed that the cycle time to pick and place three mushrooms
\nwas 20 seconds (compared to a typical human pick rate of 12 seconds (HDC 1996)).
\nThe model could in theory, be operated 24 hours a day, giving a picking strategy
\nadvantage over a current single day-shift operation. The pick efficiency rate (i.e.
\nsuccess rate) was found to be 69% and if all biological factors are eliminated (e.g.
\nelimination of air conditioning which dried out compost and fruiting bodies), the results
\nsuggest a 92% pick success rate is theoretically feasible using the model within
\noptimum environmental conditions. Additionally, 85% of these mushrooms
\nsuccessfully picked had no bruising damage; this results in an overall 78.2% success
\nrate, or 21.8% scrap rate, compared to a 5-10% scrap rate produced by human pickers
\n(Noble 2004), (Komatsu 2005), (Howard 2007).
\nThe performance of the robotic harvester was tested within a simulated commercial
\nenvironment using a discrete event simulation of a UK farm. Results of experiments
\nconducted to compare the performance of a robotic harvesting operation to the current
\nlabour intensive operation show that the system would require between 31 and 34 robot
\nharvesters to replace the current 28 humans.
\nThe initial investment cost for the proposed fully automated harvesting and growing
\nsystem, using an Automated Storage and Retrieval System, for the UK farm was found
\nto be from £3.56-3.71m. The payback period for the replacement of the 28 Flexible
\nFull Time Harvesters currently employed was found to be 8 years. The Internal Rate of
\nReturn (IRR) was found to be 4%. If the existing growing sheds and tray transport
\nsystem at the UK farm was kept in service and just the automated harvesting unit was
\nemployed, the payback period reduced to 5.5 years and the IRR was found to be 10.5%.
\nThe financial analysis provides unimpressive results; however, limitations of these
\ntraditional financial appraisal methods were identified from this work. The nonfinancial
\nbenefits provide a more compelling reason to go ahead with the proposed
\nsolution as the persistent labour supply and direct labour cost issues are currently
\nforcing the UK growers out of business.
\nThis work provides growers with a reliable automated harvesting solution and the
\nability to determine the suitability of its application within their own operations.