A program that builds the same robot year after year isn’t really a research team — it’s a maintenance crew. Team Minion has always understood this distinction. Each platform generation has been a deliberate response to what the previous one revealed, layering new capability onto hard-won operational experience.
This is the story of the Minion fleet: where each vessel came from, what it changed, and what it proved.
Minion I — The Singapore Pioneer (2014)
The original Minion was built with a single overriding goal: get to the start line and do something autonomous. For a first-generation team at a first-ever competition, that’s the right goal.
The 2014 vessel established the team’s foundational stack: ROS-based autonomy, dual electric thrusters for differential steering, a stereo camera pair, and a single 2D LIDAR. The WAM-V 16 hull provided by RobotX organizers gave the team a stable base to instrument without designing their own hull.
Minion I successfully completed the navigation channel and the acoustic pinger tasks in Singapore — a strong result for a debut. Its weaknesses (rigid sensor mounting, limited controller tuning, brittle software architecture) were precisely catalogued and became the 2016 development roadmap.
Minion II — The Hawaii Competitor (2016)
The 2016 platform kept everything that worked from Minion I and rebuilt everything that didn’t. The aluminum T-slot rail system replaced fixed sensor mounts. The station-keeping controller gained anti-windup and wind feedforward. The software architecture was reorganized around composable behavior trees.
Minion II was the first vessel to successfully complete the docking task in competition — a milestone the 2014 team had been unable to reach. It was also faster, more responsive to sensor input, and easier to configure between tasks.
Perhaps more important than its performance on the water: Minion II proved the team’s development process. Systematic post-mortems, targeted improvements, and disciplined testing produced measurable results. That process is now a template the team applies to every new generation.
Minion III — The Sensor-Heavy Platform (2018)
Between the 2016 and 2018 competitions, sensor technology matured rapidly. The team took advantage. Minion III introduced a 3D solid-state LIDAR alongside the existing 2D scanner, giving the vessel volumetric obstacle awareness for the first time. A four-camera 360-degree surround-view system replaced the forward-facing stereo pair for primary buoy detection.
The acoustic subsystem received its most significant upgrade yet: a new hydrophone array geometry — a vertical line array suspended 1.5 meters below the hull — improved elevation resolution for three-dimensional pinger localization. This was directly applicable to real-world sonar research, and the work was documented in an academic paper submitted to an ocean engineering conference.
The computing architecture also changed. Minion III introduced a dedicated GPU module for the vision pipeline, reducing perception latency from approximately 120 milliseconds to under 40 milliseconds. At the vessel’s low operating speeds this improvement was modest in absolute terms but significant for fine approach maneuvers near task structures.
Challenges Minion III Introduced
More sensors meant more integration surface area. The 360-degree camera system required careful spatial calibration to stitch adjacent views consistently — a process that had to be repeated each morning as temperature changes flexed the carbon fiber sensor mast. The GPU module drew 85 watts at full load, a non-trivial addition to the power budget that required careful management during extended run sequences.
Minion IV — Modular and Maintainable (2019)
By 2019 the team recognized a pattern: each generation had been optimized for the competition that followed it, but the accumulated complexity was beginning to slow development. Minion IV was a deliberate simplification as much as an advancement.
The 360-degree camera system was replaced by a computationally efficient stereo-plus-monocular configuration that achieved comparable detection performance with lower processing overhead. The wiring harness — which had grown organically over three generations — was completely redesigned using a backbone-and-branch architecture that allowed individual subsystems to be disconnected and tested in isolation.
Minion IV also introduced modular payload pods: self-contained enclosures housing sensors or specialized electronics that attached to the rail system and connected to the main backbone through standardized waterproof connectors. A payload pod could be removed, bench-tested, and reinstalled in under five minutes.
This modularity directly supports the team’s educational mission. Students joining the team could be handed a single payload pod to own and develop without needing to understand the entire system. Responsibility was bounded and failure-safe — a pod failure couldn’t crash the whole vessel. Learn more about how this philosophy shapes our onboarding in the Team Minion about page.
Cross-Generation Lessons
Looking across the full fleet, several principles have proven durable:
Specific failures drive the best improvements. Every meaningful upgrade traces to a specific observable failure. Abstract improvements (“make it smarter”) rarely survive the transition from proposal to implementation. Concrete failures (“the station-keeping controller oscillates in 10-knot crosswinds”) do.
Operational simplicity compounds. A system that is easy to configure, debug, and repair in the field outperforms a theoretically superior system that is difficult to operate. The rail mounting system, backbone wiring harness, and modular payload pods all represent investments in operational simplicity.
The hull is not the vessel. The WAM-V hull has been the same across all four generations. The vessel is everything above the waterline — and almost everything above the waterline has changed at least once. This distinction keeps the team from confusing platform continuity with technical stagnation.
What Comes Next
The full technical specifications for the current competition platform are documented in our boat overview, which covers sensor selection, propulsion configuration, and software architecture as they stand today.
The fleet of Minions is not a museum. Each vessel is a step — and the next step is always the most interesting one.