Building More Than Software: Griffith Students Take On a Real-World AI and Drone Integration Challenge
For five Griffith University students, the Work Integrated Learning (WIL) program offered something different from a typical university assignment. Instead of working on a hypothetical case study, they joined KJR to tackle a genuine engineering challenge: integrating live drone telemetry, AI object detection and the Android Team Awareness Kit (ATAK) into a single system that could provide real-time situational awareness for operators in the field.
Over the semester, the team designed, built and tested a working prototype while learning what it takes to develop software around live systems, unfamiliar technologies and changing requirements.
Choosing a project that felt real
The project was one of several industry placements available through Griffith University’s Work Integrated Learning program. The students selected it for one simple reason – it felt like something they could imagine working on after graduation.
"We were working with physical hardware, integrating software and trying to solve a real-world problem, that was the biggest attraction."
Several members of the team had already come across KJR through its work in AI, software quality engineering and drone technology. The opportunity to work alongside an industry partner made the project even more appealing.
The challenge
The goal was to build a system that could take information from a DJI drone, combine it with AI object detection and present that information live inside ATAK.
ATAK is a mapping platform that allows users to share locations and operational information in real time. The students’ task was to extend that capability by automatically displaying AI detections from a drone as they happened. Rather than asking someone to review drone footage after a flight, the system would identify objects during the mission and immediately display their location to both a remote operator and personnel working in the field.
To achieve this, the team integrated drone telemetry, live video streaming, AI detection, GPS calculations and mapping software into a single workflow.
Starting with more questions than answers
Like many engineering projects, the team didn’t begin with a complete solution. Their first job was simply understanding the technologies involved. They researched TAK, explored how DJI drones communicate, investigated AI detection models and worked out how information might flow between each component. Only then could they begin assigning responsibilities across the team.
Each member brought different strengths. Nick Martini coordinated the project, Lucky Mawadi focused on quality assurance, Dhruv Mehra applied his networking and security experience through a SecDevOps role, while Quinn Wathen-Dunn and Qilang He concentrated on software development and system integration.
The early weeks were spent exploring each technology, identifying dependencies and deciding who was best placed to tackle different parts of the system.
The first lesson: real systems behave differently
One of the earliest challenges arrived when the students moved from research into practical development. Their initial software had been designed using publicly available examples of drone data. Once they began working with live DJI hardware at KJR, they discovered the real telemetry behaved differently.
"The initial program didn't align with the real-world data," Nick recalled. "We had to restructure how we collected information from the drone."
It was an early reminder software built around assumptions often needs to change once it meets a real operating environment. That experience became one of the defining moments of the project.
Solving problems as they appeared
As development progressed, new challenges emerged. One of the most difficult involved the live RTSP video stream carrying imagery from the drone. If the video stream dropped out, even briefly, parts of the system would stop responding. Recovering could take several minutes, making live testing difficult and interrupting demonstrations. Rather than treating it as a one-off issue, the team investigated how the streaming pipeline behaved, refined the way the software handled interruptions and continued testing until the system became more stable.
These weren’t problems anyone could solve from lecture notes. They required experimentation, persistence and plenty of trial and error.
Learning technologies they hadn't used before
Although each student brought technical knowledge from their degree, much of the project’s technology was new. The team worked with live RTSP video streaming, DJI drone systems, ATAK, OpenTAK Server, Python, cloud infrastructure and supporting scripting tools. They also built automated processes so the platform could be launched through a simple start-up sequence rather than requiring multiple manual steps.
For Dhruv Mehra, some previous networking subjects provided useful background, but working with these technologies in a professional environment required a different level of understanding.
"The interface was totally different from how we had performed it while studying," he reflected.
Working as an engineering team
The technical work was only part of the experience. The students also learned how professional software teams organise themselves. During the first month they met regularly at Griffith University before moving much of their work to KJR’s office, supported by Microsoft Teams and regular technical discussions.
Responsibilities shifted as new problems emerged. Team members helped one another troubleshoot issues, reviewed each other’s work and adjusted plans as they learned more about the technology. It was a practical lesson in collaboration, communication and adapting to changing priorities.
KJR's role
KJR provided far more than a project brief. The team had direct access to KJR founder and Managing Director Dr Kelvin Ross, commercial drone equipment, technical documentation and an environment where ideas could be tested against live systems rather than simulations.
That support helped the students understand why engineering decisions matter in operational environments. It also exposed them to the kinds of technical challenges KJR solves for clients, where reliability, integration and software quality directly influence the outcome of a project. Rather than giving the team all the answers, Kelvin encouraged them to investigate problems, test ideas and learn through the engineering process.
Looking back
By the end of the project, the team had delivered a working prototype that connected drone telemetry, AI object detection and ATAK into a single operational workflow. Equally important was what they learned along the way.
They experienced the difference between writing code that works in theory and software that must perform reliably with live hardware. They learned unfamiliar technologies, adapted when assumptions proved wrong and discovered that engineering is often about solving the next problem rather than reaching a perfect first solution.
For KJR, projects like this are an investment in Australia’s future technology workforce. They give students the opportunity to work alongside practising engineers, contribute to genuine technical challenges and graduate with experience that simply can’t be replicated in a classroom. For the students, it was a chance to discover what professional software engineering really looks like. Oh, and their university result for this program – a 7 out of 7!
See how KJR turns real operational challenges into practical engineering solutions, whether through student placements or client projects.