I’m looking into using stereo cameras for vehicle-based road asset data collection. I wanted to ask the engineers/community here if this approach is realistic. We aren’t chasing survey-grade precision. Our target is around 20-30cm absolute accuracy for asset tagging.
- Is this accuracy achievable with a stereo setup on a moving vehicle, or is LIDAR strictly required for this range?
- What would be the minimum hardware config to hit that target? Is the internal IMU/GNSS usually enough, or would we still need an external GNSS/INS integration to get usable georeferencing?
- What does the software stack look like for this?
If anyone has done a similar setup or has sample data from a mobile run, I’d love to hear about it.
You can use a ZED X camera for this type of task. The Global Shutter sensors allow you to obtain stable images and then stable depth information even in dynamic conditions.
I recommend you consider a ZED Box Orin NX 16 GB with an integrated GNSS RTK sensor.
The ZED SDK provides GNSS data fusion capabilities: