Autonomous farming has the potential for reducing agricultural production costs, but it may become economically infeasible when 50% time or more human supervision is enforced by law. Because worldwide some jurisdictions are requiring farmers to constantly supervise crop robots during field operations, this study explores crop robot supervision levels that would be voluntarily implemented by farmers in the absence of such regulations. The core hypothesis is that human supervision only makes economic sense if a crop robot requires frequent human intervention to deal with minor performance issues, especially during difficult operations such as drilling and harvesting. By using a linear programming optimisation model developed at Harper Adams University, four scenarios at four farm sizes are generated where crop robots undergo frequent or infrequent incidents and where the human supervisor either operates crop robots from the field or from an off-farm location. Wheat production costs for these four incident scenarios are compared with two incident-free baseline scenarios that use either crop robots or conventional equipment to explore farmer supervision decisions. The results of this analysis show that if crop robots undergo infrequent incidents, a farmer will choose to supervise autonomous operations from the field for 13% of the time or remotely for 25% of the time. Conversely, if crop robots undergo frequent incidents, a farmer will choose to supervise crop robots from the field for 43% of the time on small farms but switch to 100% supervision level or to conventional equipment on larger farms, especially if operations are remotely supervised. These findings reveal crucial implications for health and safety regulators, who should strive to balance costs with farm safety benefits, and for researchers and entrepreneurs, who should prioritise the development of AI that reduces supervision needs.
The presenter wishes to warmly thank his co-authors Prof. James Lowenberg-DeBoer, Prof. Karl Behrendt and Mr. Kit Franklin at Harper Adams University (UK) for providing constant support and exceptional expertise during all phases of this study, including research conceptualisation, results validation and interpretation, and review of the final manuscript.
