Introducing ANUBIX to the agritech community — our graduation project engineered to reshape how the agricultural sector handles disease detection.
The core challenge we tackled was under research for the past 2 years in the plant diseases department at Ain Shams Faculty of Agriculture, and it simply was about timing. In agriculture, when a farmer spots a diseased crop, the infection has typically already spread, often making the treatment economically unviable. By integrating Si-Ware Systems' NIR Spectroscopy, we were able to introduce a new sensing dimension into the realm of Physical AI,detecting diseases far before visual symptoms appear.
In practice, ANUBIX operates autonomously inside a greenhouse. It maps and navigates the environment, identifies target leaves, and maneuvers a 6-DOF robotic arm to scan them using an NIR spectrometer. The system evaluates the plant's internal biochemical signature to classify Tobacco Mosaic Virus before any physical symptoms exist, immediately streaming those insights to a remote cloud dashboard.
My role as the R&D team leader, system architect and Agentic AI engineer was focused on designing isolated development stacks. bringing vision, spectroscopy, autonomous navigation, arm control and backend to life all in parallel in a timeline of 8 weeks, then weaving them together into a unified, intelligent ecosystem. I architected the distributed computing infrastructure, orchestrating a Jetson Orin Nano and Raspberry Pi cluster to handle perception, autonomous navigation, spectroscopy, arm control and agentic AI. Finally, to give the system its cognitive layer, I engineered the cloud to edge AI pipeline, deploying an LLM-driven reasoning engine via AWS and Tailscale. This agentic core empowered the robot to decompose complex agricultural tasks, make dynamic decisions, and trigger failure recovery protocols while operating within strict physical safety guardrails.
Check out more about the project on GitHub:
* https://lnkd.in/ee5rKnhJ
Building an architecture of this scale requires incredible backing. A massive thank you to our supervisor, Prof. Lamiaa Elrefaei, for championing our vision. I also want to deeply thank our technical mentors, Dr. Sherif Okda, Eng. Moez El-Massry, and Shady Reda Labib, MSc., for their crucial technical insights, and Dr. Yasser Sabry for trusting us to execute this challenge.
To my team — Hanin Sherif, Andrew Tanas, Hazem Abuelanin, Mohamed Hany and Ahmed Abdelwahed — thank you for putting in all the effort, the late nights, and the resources despite all the challenges and the days it felt impossible.
Finally, a special thanks to our corporate sponsors:
* Si-Ware Systems, without whose belief, technical mentorship and resources it wouldn’t have been possible.
* OmniLink, whose agentic platform helped build ANUBIX's agentic tools.
#Robotics #SystemArchitecture #PhysicalAI #ROS2 #EdgeComputing #ComputerEngineering #GraduationProject #Agritech
26 Julio 2026 à 16h30
Ver en LinkedIn →
