volume landmarks

This release accompanies the major revision of the VIO benchmark study comparing ESKF, Full UKF, and selective Hybrid ESKF–UKF estimator architectures. Major updates in v1.1.0 include: Reproducible controlled multi-seed benchmarking with paired random seeds and explicit failure accounting. SE(3)-aligned translational and rotational ATE and fixed-interval RPE trajectory metrics. Stereo visual-odometry frontend for real-image evaluation using rectification, ORB features, stereo reconstruction, temporal association, and PnP-RANSAC. Real stereo-image validation workflows for EuRoC MAV and TUM VI. Cross-dataset controlled evaluation on TUM VI. Matched Hybrid versus Hybrid-LinearOri orientation-correction ablation. Synthetic visual-observation noise sensitivity analysis. Real-image assumed measurement-uncertainty sensitivity analysis. Dedicated estimator-backend timing benchmark with repeated measurements and rotated estimator execution order. Updated ESKF, Full UKF, and Hybrid implementations and documentation. Self-contained publication workflows for reproducing the revised benchmark experiments. The computational timing results characterize estimator-backend processing only and should not be interpreted as end-to-end VIO or embedded hard-real-time performance. Datasets are not redistributed with this repository. EuRoC MAV and TUM VI should be obtained from their original dataset providers and organized according to the repository documentation.

matched text: “mav”

This release accompanies the major revision of the VIO benchmark study comparing ESKF, Full UKF, and selective Hybrid ESKF–UKF estimator architectures. Major updates in v1.1.0 include: Reproducible controlled multi-seed benchmarking with paired random seeds and explicit failure accounting. SE(3)-aligned translational and rotational ATE and fixed-interval RPE trajectory metrics. Stereo visual-odometry frontend for real-image evaluation using rectification, ORB features, stereo reconstruction, temporal association, and PnP-RANSAC. Real stereo-image validation workflows for EuRoC MAV and TUM VI. Cross-dataset controlled evaluation on TUM VI. Matched Hybrid versus Hybrid-LinearOri orientation-correction ablation. Synthetic visual-observation noise sensitivity analysis. Real-image assumed measurement-uncertainty sensitivity analysis. Dedicated estimator-backend timing benchmark with repeated measurements and rotated estimator execution order. Updated ESKF, Full UKF, and Hybrid implementations and documentation. Self-contained publication workflows for reproducing the revised benchmark experiments. The computational timing results characterize estimator-backend processing only and should not be interpreted as end-to-end VIO or embedded hard-real-time performance. Datasets are not redistributed with this repository. EuRoC MAV and TUM VI should be obtained from their original dataset providers and organized according to the repository documentation.

matched text: “mav”