Researchers announce watermelon picking robot

VIP ROMPER correctly identified melons ripe for picking 85% of the time.

A team of Israeli and US researchers has designed a watermelon-picking robot endowed with artificial vision to do the job of harvesting. The robot is the result of a collaboration of three Israeli Institutes of higher learning including Ben-Gurion University of the Negev, the Weizmann Institute of Science, the Agricultural Research Organization and Purdue University of the US.

Harvesting watermelon has always been relatively expensive because the process is labor-intensive.

The machine consists of a mobile platform on which are mounted an image-processing system, air blowers and a mechanical arm with a gripper attached. Tractor power pulls the platform through the field while cameras take pictures that the system analyzes. The air blowers ruffle the foliage to expose the fruit. When the harvester sights a melon bigger than a certain size -- and therefore presumed to be ripe -- it extends the gripper to grab the fruit and lift it off the ground.

Onboard software evaluates the image's shape, brightness, and texture to locate the melons. Knives connected to the gripper slash the stalk, and the gripper places the melon on a conveyor belt.

The harvester, named VIP ROMPER, guides itself down rows of maturing melon plants with only occasional human steering corrections.

In field tests, VIP ROMPER correctly identified melons ripe for picking 85% of the time. Prof. Yael Edan of Ben Gurion University said that she estimates a two-armed version could attain a picking rate of one and a half seconds per melon.

Watermelon is grown in 90 countries with worldwide production exceeding 50 billion pounds per year.

Ben-Gurion University said that the robot is now being commercialized.

Published by Globes [online] - www.globes.co.il - on 11 September 2003

5 Comments
View comments in rows
Update by email about comments talkback
POST
Comments
Twitter Facebook Linkedin RSS Newsletters גלובס Israel Business Conference 2018