Robot harvesters may be coming for table grapes. They use cameras with artificial intelligence to select and cut bunches based on color and size. One example is from Tortuga AgTech. While there may be some limitations to the existing technology, Sensonomic is excited to use it to supercharge our yield estimation and harvest planning algorithms.
First, let’s talk about the drawbacks. An obvious one is that the machines are slow. A pair of people can harvest up to 180 boxes (8.2 kg/18 lb) per day, which I’m guessing is much faster. To make up for it, the robots can harvest 24 hours a day.
Harvest rate also depends on whether trimming and packing is done in the field. The robot does neither, so grapes must be moved to a central packhouse. This can take a long time as the machine can only carry three boxes.
Here is where four-wheeled, GPS-enabled carrier robots from Burro could come in. They could ferry boxes between the harvesters and the packhouse. Something will need to be done about softening the impact when transferring the fruit to the boxes to avoid damage.
Cost is going to be another factor. It’s no surprise that robot harvesting began in strawberries, where margins are likely higher. How low will costs need to go? California grape pickers get paid $21/hour. Using a harvest rate of 180 boxes per eight hour day, the cost of harvest is $2/box. To outcompete the human harvesters, a robot is going to have to cost less.
Viewed another way, doing a very basic extrapolation of the 24-hour harvest rate from the video (260 boxes/day), the robot is going to have to cost less than $520/day. These numbers may be nonsense, but they indicate the type of calculations needed to compare costs.
There may be an additional cost of restructuring the canopy. Vines may need to be retrained for the harvester to work efficiently. Elsewhere growers have transformed apple orchards into fruit walls for robot picking or olive groves into hedgerows for mechanical harvesting.
Notwithstanding their costs and constraints, we at Sensonomic welcome our new robot overlords. For one thing, we can feed their automated calculations of bunch and berry size into our algorithms for predicting final volumes at various stages of development.
What’s perhaps even more promising is that we can guide the teams of harvester and carrier robots with our harvest planning algorithms. Using our predictions of sugar accumulation, we can instruct the harvest teams on which blocks to go to at which times to ensure a continuous flow of high quality fruit to the packhouse. Field data can be used to update plans in real time.
At the end of the day it doesn’t really matter if the communication is with people or robots. Sensonomic’s algorithms can be used in either case to generate accurate yield predictions and execute an efficient harvest. Achieving these objectives will reduce costs and extract more value from harvested grapes.
Sign up to our mailing list: https://eepurl.com/gaXLkr
