A maritime system, not a boat.
Launched in 2012 by the U.S. Office of Naval Research with Singapore's Future Systems and Technology Directorate, the Maritime RobotX Challenge brings teams from across the Pacific Rim to build autonomous surface vehicles and the sensing that lets them work in a live maritime environment.
Entries are judged as an Autonomous Maritime System rather than as a single hull. A surface vehicle carries the mission, and an aerial vehicle launches from it to reach what the boat cannot. Three of the course's tasks exist only if that pairing works.
USV 1.0
Surface mission platform. WAM-V hull, differential drive on two Navy 3.0 Evo propellers, ZED 2i perception, a hydrophone, and the landing dock the aircraft returns to.
UAV 1.0
Aerial asset, launched from the boat. X650 quadrotor with a 5-DOF arm for the tin transfer and a 64 MP camera for search and report.
What the campaign demanded.
RobotX scores a system, not a vehicle. These are the problems that sat between the two machines rather than inside either one.
Two vehicles, one system
The boat is the aircraft's transport, launch pad and landing deck. It carries UAV 1.0 through transit on a dock that clamps its legs with four stepper-driven rods, releases it over the course, and takes it back aboard while under way. Replenishment, search and recovery all collapse the moment that pairing stops working.
Reporting as a task in its own right
The first task on the course is not navigation, it is telling the organisers what you are doing. The system opens every run by transmitting a heartbeat to the Technical Director network and keeps reporting through the tasks that follow. That runs over a wired link at the course tent and the team's own long-range maritime radio out to the vehicle, so status leaves the water continuously.
One reading, four tasks
Scan the Code shows a three-colour sequence once, and its colours decide which way to run the path, which creature to circle, which bay to dock in and which tin to collect. Miss it and nothing fails outright. Four separate tasks simply execute correctly against the wrong target.
A run with no operator in it
Qualifying allows a vehicle to be positioned by remote control. Semi-finals and finals do not: the system enters through a gate, runs the course autonomously end to end, and has to carry information gathered in one task into the execution of another. Autonomy stops being a feature and becomes the entry condition.
One screen for the whole system
Rebuilding the ground station was an explicit design objective, not a convenience. A single-frame interface carries every vehicle status alongside a video feed of the aircraft's onboard computer, so a fault can be located while the run is still going rather than diagnosed afterwards from a log.
The campaign on film.
RobotX asks every team for a video introducing themselves and their system, hosted publicly.
The 2024 team.
The engineers who took two vehicles to Sarasota, three months after taking the project over.
