University students harness UR robots and AI for agriculture

This summer, a group of ambitious students from California State University, Monterey Bay (CSUMB) embarked on a mission to tackle one of agriculture’s most pressing challenges: labor-intensive fruit harvesting. Armed with a Universal Robots collaborative robot, the latest PolyScope X software, and the AI Accelerator, these students turned their classroom knowledge into a real-world proof of concept – demonstrating how automation and AI can revolutionize the agricultural industry.

The students’ project aimed to create a robotic system capable of identifying and sorting fruit using AI-powered vision. Their goal? To explore how automation could ease the physical burden on agricultural workers and improve food availability by increasing harvesting efficiency.

For Estrella Chavez Gaeta, a mechanical engineering student who currently lives on a vineyard, the project was deeply personal. “Agriculture is often overlooked when it comes to automation,” she shared. “But these are some of the most labor-intensive jobs. Our project wasn’t about replacing people – it was about making their work safer and more sustainable.”

For many of the students, this was their first hands-on experience with robotics – and Universal Robots made it a memorable one.

“I had never worked with robotics before,” said Josh Williams, a mathematics and economics student. “But from the start, Universal Robots stood out. The branding was sleek, the interface was intuitive, and the support was phenomenal.”

Cable Bronte, a mechatronics student, echoed the sentiment. “I worked a lot with the teach pendant and PolyScope X. It was straightforward and intuitive – even when we had to switch from PolyScope 5 to X mid-project. The UR Academy tutorials were helpful, and the new interface felt modern and refined. Additionally, the URScript Documentation available online allowed us to integrate custom code with confidence.”

The AI Accelerator, offered in partnership with NVIDIA, embeds and seamlessly integrates the NVIDIA Jetson Orin AGX 64GB compute box, providing powerful on-device AI processing capabilities. This advanced hardware works directly with PolyScope X, making it simple to train and deploy models – even for first-time learners. Devyn Korman, lead labelling for the project, noted, “The interfaces were clean and simple to use. Once the images were labelled it was simple to import into PolyScope where we had a working model within 20 minutes.”

The team faced their share of hurdles – from addressing environmental challenges like fruit degradation affecting model accuracy to navigating the crossroads of robotics and AI - Physical AI - for the first time. But problem solving these challenges only deepened their learning outcomes.

The students envision a future where UR robots are deployed directly in the field – picking fruit, identifying ripeness and even detecting defects. “We want to build stronger models, collect more data and eventually take this from the lab to the farm,” said Cable.

Their project culminated in a public symposium where they presented their work to faculty, students and community members. The response was overwhelmingly positive, and the team even won an award for Outstanding Poster and Presentation.

Josh is studying Mathematics and Economics at CSUMB and plans to transfer to Stanford to pursue a PhD in Finance. He led the AI integration and terminal programming efforts and was instrumental in getting the system up and running.

Cable is a junior in the Mechatronics program at California State University, Monterey Bay (CSUMB). As the team’s planner and lead on teach pendant programming, Cable played a key role in integrating the robotic arm and navigating the transition to PolyScope X.

Estrella began her journey at Gavilan College and recently transferred to Santa Clara University to pursue a degree in Mechanical Engineering. Recognizing the significance of the agriculture industry on the local and national economies, Estrella brought a personal connection to the project and passion for applying robotics to solve real-world farming challenges. She contributed to the project by supporting end effector 3D modeling, 3D printing, & integration, and assisting with image labeling for AI training.

Devyn is a second-year Mechatronics student at CSUMB. This internship marked his first professional experience, where he contributed to image labeling and AI model training. Devyn found the process surprisingly intuitive and was excited by the hands-on learning.

The team gratefully acknowledges the mentorship and guidance of Dr. Jeorge Hurtarte, as well as the support, time, and expertise of Zafar Kamil at Universal Robots and Javier Coyte at CSUMB.