Background
A flight-style ride differs from a roller coaster in almost every dimension: gondolas move relatively freely, every ride takes a different path, and the experience lives in the perspective of the individual riders. A single fixed on-ride camera at entry/exit doesn't capture that.
Plutos wanted a per-gondola video experience from multiple cameras at once. That introduced three hard technical constraints: reliable triggering at exactly the moment the ride starts; real-time local capture from many cameras per gondola; and robust transfer from the gondolas to a central server at ground level — under real-world park conditions with vibration, weather, and staff who can't be expected to debug SSH sessions.
Solution
We built the system layer by layer.
- Custom carrier boards for Raspberry Pi Compute Modules. Each gondola gets its own compute platform on a Compute Module; the custom carrier board hooks up cameras, sensors and power in a form factor that fits the gondola and survives shock and vibration.
- Eight cameras per gondola over a mini-network. The video and image capture per gondola lands on a central Pi that buffers everything in sync. Accelerometers feed extra motion data for downstream analysis.
- Triggering via an Arduino on the ride's PLC. The Arduino reads the ride-start signal directly from the PLC, interprets it, and forwards it to the Raspberry Pi modules. Capture is locked to the actual ride event, not an approximation.
- Transfer and operations. As soon as a gondola comes to rest, the structured transfer to the ground-side server kicks off. Dashboards expose capture and transfer state per gondola, so operators can see at a glance — without any Linux knowledge — that the system is running.
Outcome
What started as "a cool idea for the ride" turned into a distributed real-time video system that gets from trigger to finished per-gondola video within a minute. Complex hardware integration, a park environment full of real-world noise, and ultimately a system that adds a whole experience layer without disrupting ride operations.
Why this fits us
A cloud-first stack, an off-the-shelf SaaS or a slide-deck architecture is useless here. This needs people who move between PLC programming, Compute Module pinouts, Python real-time code and park-operations reality — and who build a system still running three seasons later, without park staff having to take an IT course.