Data acquisition and visualisation for autonomous vehicles
A high-throughput application for capturing, streaming, and visualising sensor data from autonomous vehicles during road testing.
- Streams
- Many, synchronised
- Mode
- Live + record
- Challenge
- Record and display many synchronised sensor streams from a moving vehicle at full rate, without dropping frames or losing time alignment.
- Approach
- A native C++ application built on a Qt interface and a GStreamer media backend, engineered for sustained throughput and low-overhead visualisation.
- Outcome
- A tool the test team uses to watch and review vehicle data live, with recordings that stay synchronised for later analysis.
Stack
- C++
- Qt
- GStreamer
- Real-time systems
Background
An autonomous mobility team was limited by its data tooling. During road tests they needed to see multiple sensor feeds live and capture them at full rate, but existing tools dropped frames and lost synchronisation between streams — which made post-test analysis unreliable.
What we did
We wrote a native application in C++ with a Qt front end and a GStreamer backend. The design prioritised sustained throughput: zero-copy paths where possible, careful buffering, and a visualisation layer light enough to run alongside recording without stealing headroom from capture. Timestamps were carried end to end so recordings stayed aligned.
Result
The test team now reviews vehicle data as it is produced and trusts the recordings afterward. The tool became part of their standard test workflow.
Engagement details are adapted and figures are confirmed with the client before publication.