Skip to content

Python Package Usage

In addition to the GUI, trajectopy can also be used as a Python package. The package provides a set of functions to evaluate and visualize the quality of trajectory. Trajectopy is designed to be easy to use and to provide a high level of customization.

Absolute Trajectory Error (ATE)

Computing the Absolute Trajectory Error (ATE) between two trajectories is a common evaluation metric in the field of visual odometry and SLAM. The ATE is calculated by aligning the two trajectories and then computing the pose-wise difference between the aligned trajectories.

import trajectopy as tpy

gt_traj = tpy.Trajectory.from_file("./example_data/KITTI_gt.traj")
est_traj = tpy.Trajectory.from_file("./example_data/KITTI_ORB.traj")

ate_result = tpy.ate(trajectory_gt=gt_traj, trajectory_est=est_traj)

Relative Pose Error (RPE)

The Relative Pose Error (RPE) is another common evaluation metric. The RPE measures the relative pose difference between two trajectories at different time or distance intervals.

import trajectopy as tpy

gt_traj = tpy.Trajectory.from_file("./example_data/KITTI_gt.traj")
est_traj = tpy.Trajectory.from_file("./example_data/KITTI_ORB.traj")

rpe_result = tpy.rpe(trajectory_gt=gt_traj, trajectory_est=est_traj)

Single Trajectory Evaluation Report

The results of a single trajectory comparison can be visualized in a HTML report.

# Import
gt_traj = tpy.Trajectory.from_file("./example_data/KITTI_gt.traj")
est_traj = tpy.Trajectory.from_file("./example_data/KITTI_ORB.traj")

# Evaluation
ate_result = tpy.ate(trajectory_gt=gt_traj, trajectory_est=est_traj)
rpe_result = tpy.rpe(trajectory_gt=gt_traj, trajectory_est=est_traj)

# Report
report_settings = tpy.ReportSettings(ate_unit_is_mm=False, scatter_marker_size=8)
report = tpy.create_deviation_report(ate_result=ate_result, rpe_result=rpe_result, report_settings=report_settings)
tpy.show_report(report_text=report, filepath="reports/report.html")

Multi Trajectory Evaluation Report

Multiple trajectory comparisons can be visualized in a single HTML report.

import trajectopy as tpy

# Import
gt_traj = tpy.Trajectory.from_file("./example_data/KITTI_gt.traj")
est_traj_1 = tpy.Trajectory.from_file("./example_data/KITTI_ORB.traj")
est_traj_2 = tpy.Trajectory.from_file("./example_data/KITTI_SPTAM.traj")

# Evaluation 1
ate_result_1 = tpy.ate(trajectory_gt=gt_traj, trajectory_est=est_traj_1)
rpe_result_1 = tpy.rpe(trajectory_gt=gt_traj, trajectory_est=est_traj_1)

# Evaluation 2
ate_result_2 = tpy.ate(trajectory_gt=gt_traj, trajectory_est=est_traj_2)
rpe_result_2 = tpy.rpe(trajectory_gt=gt_traj, trajectory_est=est_traj_2)

# Report
multi_report = tpy.create_deviation_report(
    ate_result=[ate_result_1, ate_result_2], rpe_result=[rpe_result_1, rpe_result_2]
)
tpy.show_report(report_text=multi_report, filepath="reports/report.html")

Matplotlib Plotting

In addition to the HTML-based plotting, trajectopy also supports Matplotlib-based plotting. In the below example, all available plotting functions are demonstrated.

import matplotlib.pyplot as plt
import trajectopy as tpy

# importing
gt = tpy.Trajectory.from_file("./example_data/KITTI_gt.traj")
orb = tpy.Trajectory.from_file("./example_data/KITTI_ORB.traj")
sptam = tpy.Trajectory.from_file("./example_data/KITTI_SPTAM.traj")
tpy.plot_trajectories([gt, orb])

# evaluation
alignment = tpy.estimate_alignment(gt, orb)
ate_result_orb = tpy.ate(gt, orb)
ate_result_sptam = tpy.ate(gt, sptam)
rpe_result = tpy.rpe(gt, orb)

# alignment parameter plots
tpy.plot_covariance_heatmap(alignment.position_parameters)
tpy.plot_correlation_heatmap(alignment.position_parameters)

# ATE plots
tpy.plot_compact_ate_hist(ate_result_orb)
tpy.plot_ate([ate_result_orb, ate_result_sptam])
tpy.plot_ate_bars([ate_result_orb, ate_result_sptam], mode="positions")
tpy.plot_ate_bars([ate_result_orb, ate_result_sptam], mode="rotations")
tpy.plot_ate_edf(ate_result_orb)
tpy.scatter_ate(ate_result_orb)

# RPE plot
tpy.plot_rpe(rpe_result)

plt.show()