A Visual-Inertial Sensor Calibration Library

For the past couple months, I’ve been working on an open source library that I wish I had while working on sensor calibration problems for the past 6 years at Amazon/Google. I’m happy to share its initial release today!

Calico is a lightweight visual-inertial calibration library that allows for rapid problem construction, debugging, and tool creation.

Unlike other codebases that are limited to standalone calibration tools for specific hardware setups, Calico is a flexible library that enables building and modifying custom tools to suit your hardware requirements. If your calibration problem aligns with the Calico geometry paradigm, there is no need to modify the underlying optimization.

Calico features:

  • Sensor intrinsics, extrinsics, and latency estimation.
  • The ability to add multiple sensors.
  • Measurement outlier tagging and exclusion.
  • Per-sensor robustifier kernels.
  • Ability to create custom sensor models.
  • Flexible auto-differentiable cost functors.

This library is released under the MIT license. Businesses and academics alike are free to copy, modify, and redistribute as they see fit.