The Agricultural Revolution begins with the help of AI and Robotics
Almost all the process of agriculture has been automated to some extent. Harvesting has also been done mechanically for decades in crops such as potatoes and wheat. But, for crops like iceberg lettuce, tomatoes, strawberry, and other delicate fruits and vegetables, it is still challenging. Currently, available types of robots can’t be employed in the field for such delicate tasks. These robots have lots of limitations to be accepted for commercial applications.
Whether it’s human or robot, to harvest, one of the important things to be considered is whether the fruit or vegetables is ripe or not. And to check for the harvest compliance (ripeness, size, infection), a robot will need a visual aid to identify and classify the issues. If the harvested crops don’t meet the market standards, it will be a complete waste and loss for the farmer. A lot of techniques and lab tests have been performed and with the advent of deep learning the use of convolutional neural networks (CNNs) is also being tried.
Grapes have been detected with Canny Edge filters, using decision trees as the classification mechanism (Berenstein, Shahar, Shapiro, & Edan, 2010). Cucumbers were detected using NIR photography at two positions 5 cm apart, to give stereoscopic depth information (Van Henten et al., 2006) and classified for maturity by estimating their weight from the perceived volume (Van Henten et al., 2002). In the recent experiment of detection of broccoli head using an RGB-D sensor has been tested but, in the outdoor condition, it was not been able to avoid interference of the light. Like the examples mentioned above, there are a number of autonomous harvesting systems but harvesting is a challenging task. The harvesting procedure differs depending upon the crops.
A scientist from the University of Cambridge has developed a harvesting robot Vegebot which uses machine learning techniques. AI in cooperation with visual solution help robot to identify and harvest challenging agricultural crops like Iceberg lettuce. Machine learning helps in improving the algorithm with experience without requiring to code it again.
The robot prototype was initially tested and trained to recognize and harvest iceberg lettuce in a lab setting. Then the experiment was performed in the field condition and found successful. Currently, the prototype is not even close to the human worker but, it is just an example that robots can be used for harvesting even a mechanically challenged crop using AI. The experiment has opened the vast market of harvesting delicate crops because if the lettuce can be harvested using a robot then it is possible to make a robot that can harvest other crops also. But, there is considerable research required still to make it commercially possible.
A similar approach has been shown by Fieldwork Robotics. The company has also tested the prototype of the raspberry harvesting robot system. Fieldwork Robotics is a spinout company from the University of Plymouth. Agrobot is also developing a robotic system for harvesting strawberry. The robot will be using real-time artificial intelligence for determining the ripe strawberry and choose which ones to harvest. The robot is integrated with color and infrared depth sensors to select for harvesting.
Crops harvesting depends upon the various factor. The system design and function completely depend upon the condition and type of the crop. Vegebot is developed on the UR10 collaborative robot with two main components – the vision system and cutting system. The overhead camera takes images and identifies each lettuce and differentiates which one has to be harvested. The second camera on the cutting system helps in cutting without damaging the lettuce. The UR cobot helps to measure the torque and force information.
The approach from the different companies and research institute has made it clear that there is a requirement of the robotic system to overcome the shortage of labor in the agriculture industry, especially in the developed countries. This experiment has helped to overcome the few barriers related to the acceptance of robotics in the agriculture industry but still, there are a lot of limitations to overcome. The first and foremost challenge is the speed of the robot and its ability to function in a variety of field types, sizes, and crop layouts. The robot in the future will also be able to collect the data which can be analyzed later on for the further improvement of the efficiency and profit. Advancements in artificial intelligence have made this stage possible and it can be expected that these limitations would also be overcome in the near future.
To deep dive and stay continuously updated about the most recent global innovations in Artificial Intelligence / Robotics and learn more about applications in your industry, test drive WhatNext now!