uzh-rpg/rpg_trajectory_evaluation

Health v2 · maintenance only

Adoption not part of the score

1212 stars · 374 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

A Python toolbox for quantitatively evaluating visual(-inertial) odometry trajectories, supporting multiple alignment methods and standard error metrics like ATE and relative error. It can analyze single trajectory estimates or compare multiple algorithms across datasets, producing paper-ready plots and tables.

Use cases

• evaluate visual odometry trajectory accuracy against ground truth
• compute absolute trajectory error for SLAM output
• compare VIO algorithms across multiple datasets
• generate plots and tables of odometry errors for a paper
• align estimated trajectories with rigid-body or similarity transforms
• run KITTI-style relative pose error evaluation

When to choose

• you need standardized ATE/RE metrics for VO/VIO research
• you want reproducible, paper-ready trajectory evaluation plots and tables
• you are benchmarking multiple odometry algorithms on many datasets
• you work in a ROS environment and want an evaluation package

When to avoid

• you need Python 3 support (only Python 2 is supported)
• you need real-time or online trajectory evaluation
• your task is general SLAM mapping quality rather than pose trajectory error
• you need evaluation of non-pose sensor outputs

Facets

library · maturity maintenance

benchmarking data-visualization math robotics computer-vision developer-tools python cli slam visual-odometry trajectory-evaluation ate relative-error ros research-tool algorithms linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="uzh-rpg/rpg_trajectory_evaluation")