Not only the robot is changing, the crop system must evolve as well

Robotisation in agrifood is often discussed as the introduction of a new machine in the production process. In fruit production, it is becoming clear that the biggest shift often lies outside the machine itself. If a robot needs to recognise and pick apples, it helps when the orchard is structured in a predictable way and supported by digital tools. This raises a broader question for growers, technology companies and researchers. Do we keep adapting robots to existing crop systems, or will crop systems increasingly adapt to automation?

According to Han Smits, director at Munckhof Fruit Tech Innovators, the tree must adapt to picking robots. In practice this mainly means reducing variation. Robots perform more reliably when fruit positions, branch structures and row layouts are more predictable. This reduces the need for the technology to estimate and allows it to operate more consistently.

Munckhof is developing an apple picking robot based on the Pluk O Track platform, which is already widely used by growers. The current technology can pick roughly one apple every 6.5 seconds and harvests about 50 percent of the apples on a tree. The goal is to reach 3 to 4 seconds per apple with a picking rate of 70 percent within three to four years. In practice, it is precisely this combination that determines whether investments become viable and whether the system is reliable enough for daily operations.

To improve robot performance, attention is shifting towards autonomous blossom thinning at tree level. Digital thinning can help create a simpler fruit structure, ideally one apple per branch. This makes the orchard more uniform and brings the transition from prototype to real practice closer. Experiments are currently running at FRC Randwijk and in South Tyrol to test this approach under real cultivation conditions.