System Overview
A collapsible VTOL tailsitter, built to meet hard constraints
Our aircraft takes off and lands vertically, transitions to efficient forward flight for the search, and folds down small enough to carry on. Every subsystem is scoped against the SUAS 2027 ruleset below.
Hardware Architecture
Subsystem breakdown
- Airframe (collapsible, VTOL tailsitter)
- Propulsion
- Onboard compute
- Imaging
- Communications (air + ground)
- Battery / power
- Payload delivery mechanism
Software Architecture
Flight software stack
A companion computer runs ROS 2 alongside the PX4 autopilot, bridging sensors and mission logic to flight control, and the OpenHD link to the ground station. Gazebo runs the identical stack in simulation so autonomy and control code gets flight-tested before it ever touches hardware.
PX4 Autopilot
Runs on the flight controller. Executes low-level stabilization, the VTOL hover-to-cruise transition, and waypoint navigation from commands handed down over MAVLink.
View source / docsROS 2
The companion computer's nervous system: connects perception, mission logic, and payload-control nodes over pub/sub topics so subsystems can be developed and tested independently.
View source / docsMAVROS / MAVLink
Bridges ROS 2 and PX4, translating mission commands, parameters, and telemetry between the companion computer and the flight controller.
View source / docsOpenHD
Long-range digital link carrying live video and telemetry down to the ground station, and command uplink back to the aircraft.
View source / docsGazebo
Software-in-the-loop sim that feeds synthetic sensor data into the same ROS 2 / PX4 stack used in flight, so autonomy and control logic are validated before every real flight test.
View source / docsQGroundControl + custom dashboard
Operator-facing mission planning, live telemetry, and video display, plus manual override. This is what the two operators actually watch and fly from.
View source / docsDev environment: PX4, ROS 2, and Gazebo all run containerized in Docker, so the simulation and flight-control stack is reproducible across every team laptop.
Mission Autonomy Pipeline
How the vehicle flies the mission
The control code steps through four phases per flight, pulled straight from the team's planning whiteboard.
Laps
Fly the required waypoint laps to open the mission.
Inputs
- Waypoint array: latitude, longitude, MSL
- Live GPS position: latitude, longitude, MSL
Workflow
- Set the next waypoint in the array as the active goal.
- Yield (ROS job) until the goal is reached.
- Loop through the waypoint array for the required number of laps.
Risk Mapping
Cover the mapping bounds with a snake search to build the risk map.
Inputs
- Mapping bounds: bottom-left & top-right coordinates
- Coverage area per imaging pass
Workflow
- Fly along +x̂ until out of bounds.
- Step +1 in y.
- Fly along −x̂ until out of bounds.
- Step +1 in y again, and repeat until the bounds are fully covered.
Search, Detect & Deploy
Find the tent and the mannequin, then deliver the matched payload to each.
Inputs
- GPS position, wind speed, drone speed, altitude
- Payload mass & aerodynamics
Workflow
- Run detection on the mapped imagery to locate the tent and the person (ML-based).
- Convert detected pixel coordinates to GPS coordinates and send the target to PX4.
- Calculate the entry point and drop point from GPS, wind, speed, altitude, and payload data.
- Fly to the entry point, then the drop point (ROS job for each).
- Release the payload.
Exact release trigger (which completion event fires the drop) is still being pinned down.
Landing
Return and land using a pattern authored ahead of time.
Workflow
- Upload a landing pattern built in QGroundControl.
- Command PX4 to land.
SUAS 2027 Requirements
Key vehicle requirements
| Requirement | Target |
|---|---|
| Target all-up weight | ≤ 15 lb (max 35 lb) |
| Stowed envelope | ≤ 22 x 14 x 9 in (carry-on) |
| Assembly time | ≤ 180 s, no tools, ≤ 2 people |
| Configuration | VTOL tailsitter, collapsible frame |
| Cruise airspeed | ≥ 25 m/s at 110°F |
| Operating altitude | 150–400 ft AGL |
| Climb / descent angle | ≥ 20° |
| Wind tolerance | 14 mph sustained, gusts to 22 mph |
| Payloads | Strobe beacon → tent, water bottle → mannequin |
| Delivery accuracy | ≤ 50 ft (15.2 m) from target |
| Nav. accuracy | ≤ 5 m horizontal error |
| Link reliability | No outage > 5 s anywhere in flight boundary |
Full traceability: 91 competition-level requirements flow into 28 system-level and 170 subsystem-level requirements, tracked internally and refined at each design review.
Mission Timeline
Concept of operations
| Phase | Mission window |
|---|---|
| Setup + Climb | 0–5 min |
| 10 Waypoint Laps | 5–20 min |
| Risk Mapping | 20–30 min |
| Search, Detect, Deploy | 30–35 min |
| Return + Land | 35–40 min |
| Data Transfer | 40–45 min |
Scoring
How the mission is scored
| Task | Points |
|---|---|
| Flight Endurance | 250 |
| Efficient Operators | 200 |
| Design for Rapid Response | 200 |
| Search, Detect, and Deploy | 200 |
| Risk Mapping | 150 |
| Mission total | 1000 |
| Technical Design Report + website | 250 |
Design Review Milestones
SRR → PDR → CDR → FRR
September 25, 2026
Mission needs, feasibility, requirements baseline
November 6, 2026
Architecture review, subsystem concept validation
February 19, 2027
Design maturity, pre-integration readiness
March 26, 2027
Flight readiness, HITL sign-off, go/no-go
