Open Source

RoboMetrics

RoboMetrics is a lightweight Python library for robotics metrics.

It was born from a problem I repeatedly encountered while building autonomous driving systems at Minus Zero: the same trajectory, prediction, safety, and motion-quality metrics were being reimplemented across projects with slightly different assumptions, edge cases, and APIs.

Rather than rewriting them again, I packaged them into a small, reusable library.

RoboMetrics focuses on:

• Trajectory metrics (ADE, FDE, path length, curvature)

• Prediction metrics (MinADE, MinFDE, miss rate)

• Safety metrics (collision rate, minimum distance)

• Motion quality and comfort metrics (acceleration, jerk, smoothness)

• Physical consistency checks

No simulators. No cloud services. No heavyweight dependencies.

pip install robometrics

Just a simple library that you can install, import, and use inside your robotics stack.

from robometrics import ade, fde, jerk_cost

score = ade(prediction, ground_truth)
comfort = jerk_cost(trajectory, dt=0.1)