Skip to main content Skip to main navigation Skip to footer content

California Agricultural Technology Institute

Aiming for automation in tree fruit harvest

Faculty member leads student team in exploration of robot technology for citrus harvest

A Fresno State specialist in automated systems is searching for ways to apply robotic harvesting technology to the tree fruit industry in California.

Professor Nitaigour Prem Mahalik

Industrial technology Professor Nitaigour Prem Mahalik has been working with automated control and processing systems for many years. Recently, in a project conducted with support from the California State University Agricultural Research Institute (ARI), Mahalik led a team of Fresno State students in developing and testing robotic systems that could be used to harvest a citrus crop.

Global competition is pressing American farmers on many fronts, Mahalik learned from a literature search in preparation for the project. Rising costs of hand labor, combined with mounting regulatory hurdles, have created new incentives for the tree fruit industry to develop alternate harvest methods that require advanced technology, he said.

Current application of robotic systems lies mainly in assembly line and welding operations in the automotive and electronics industries, Mahalik explained. Tree fruit harvesting requires an entirely different approach to automation and system integration since there are so many more variables involved. For example, you can t place a row of citrus trees on a conveyor belt with each fruit in exactly the same location to be mechanically plucked by a robotic arm simply repeating the same motion endlessly.

Prototype robotic technology for citrus harvesting features the use of mechanical arms picking fruit while guided by image technology, which uses a camera to identity the fruit and then a guidance system to direct the arm to the individual fruit to be picked.

This process is complex and time-consuming, Mahalik said. There are many individual systems that are used in this process, and they are mutually exclusive of each other. The challenge is how to integrate all those components into one system, he said.

The $25,000 seed grant Mahalik obtained from ARI was used to support a team of students focusing on several aspects of robotic navigation and operation. They used $10,000 to purchase a small robot from a commercial company. The remaining funds supported purchase of additional parts and equipment, programming software, and for providing stipends for two students working on graduate degrees.

In all, seven students in Fresno State s industrial technology program worked on the project, completing project research, senior problems, and independent study assignments. Learning activities focused on robot navigational principles (Q-learning), radio frequency testing and analysis, algorithm development, and conceptual end-effector (the device that actually plucks the fruit from the tree) design.

Through their work, the students learned the underlying complexity of a multi-disciplinary system that integrates mechanical, electronics, control, and communication fundamentals, Mahalik said. The knowledge the students gained can be applied in real-world design, whether it is for an orange harvesting robot, a sprinkler-based frost control system, or any other automated control system, he said.

Despite the advances in robotic arm technology, this approach may not be the most cost-efficient method for tree fruit harvesting, Mahalik said.

We ve studied what can be done. The question is whether this system is viable, he said. Perhaps there is a new way of designing an end-effector.

In recent discussions with research colleagues, Mahalik has proposed the idea of a matrix-based system that would contain a framework of vacuum cylinders. The framework, housed on a moving vehicle, could approach a tree, and the cylinders all functioning simultaneously would suck individual fruit pieces off the branches using vacuum pressure from cuboid pixels, defined by GPS technology, layer by layer.

Such a system would eliminate the need for image processing, which is computationally intensive (hence slower), an expensive and complex technology required for any type of mechanical arm picking, he said.

For more information on robotic technology for citrus and tree fruit harvesting, contact Mahalik at