MichaelGrupp/evo
Evo
evo es un framework de Python para la evaluación de algoritmos SLAM, odometría de robots y datos de trayectoria. Sirve como biblioteca de análisis para medir la deriva y la precisión calculando errores de pose absolutos y relativos entre las rutas estimadas y las referencias de verdad fundamental (ground truth).
El proyecto proporciona un marco de alineación geométrica para corregir la rotación, traslación y escala entre trayectorias espaciales, asegurando una medición de error consistente. Incluye herramientas especializadas para el análisis de deriva de odometría y el procesamiento de datos robóticos, incluida la capacidad de extraer información de trayectoria de archivos ROS bagfiles.
El software cubre una amplia gama de capacidades, incluida la visualización de trayectorias en 2D y 3D con soporte para mosaicos de mapas geográficos y superposiciones de mapas ROS. La funcionalidad adicional incluye sincronización de marcas de tiempo, transformaciones espaciales y la capacidad de filtrar o exportar datos de trayectoria en varios formatos estándar de la industria.
Features
- SLAM and Odometry - Provides a comprehensive framework for measuring the precision and accumulated drift of SLAM and odometry systems.
- Analysis Libraries - Serves as a specialized library for measuring drift and precision in robot odometry by comparing estimated paths against ground truth.
- Trajectory Accuracy Evaluation - Quantifies localization precision by computing absolute and relative pose errors (ATE and RPE) against ground truth.
- Ground Truth Comparisons - Calculates absolute and relative pose errors by comparing estimated trajectories against recorded ground-truth data.
- Trajectory Error Quantification - Compare estimated poses against a reference to evaluate global consistency and the overall accuracy of a trajectory.
- Robotics Format Importers - Ships a modular input system with dedicated drivers to parse diverse robotics trajectory standards and ROS bagfiles.
- Spatial Trajectory Plotting - Renders high-performance 2D and 3D spatial views of poses and coordinates for trajectory analysis.
- Affine Matrix Transformations - Applies 3x3 or 4x4 affine matrices to map and shift coordinates for 3D translation and rotation.
- Evaluation Tools - Provides a framework for calculating absolute and relative pose errors to quantify SLAM algorithm accuracy.
- Trajectory Visualizers - Plots trajectory poses in 2D or 3D views and generates charts for coordinates, angles, and velocities.
- Evaluation Frameworks - Provides tools to quantify SLAM accuracy by computing absolute and relative pose errors against ground truth.
- Relative Pose Error Metrics - Compare the difference between consecutive or overlapping pose deltas to track local accuracy and drift over time.
- Coordinate Frame Transformations - Implements a pipeline for applying rotation and translation matrices to shift trajectory data between local and global spatial frames.
- Trajectory Alignment Tools - Provides algorithms to transform and align estimated trajectories with ground truth for direct geometric comparison.
- Trajectory Error Quantification - Quantifies trajectory accuracy by computing root mean square error and median translation differences against ground truth.
- Planar Pose Projections - Flattens 3D trajectory poses into specified 2D planes such as xy, xz, or yz for simplified analysis.
- Robotic Trajectory Standardizers - Reads trajectory and ground truth data from common robotics dataset standards for subsequent evaluation.
- Processing Utilities - Offers utilities for importing, filtering, and aligning trajectory data from ROS bagfiles and other industry-standard formats.
- ROS Libraries and Tools - Provides utilities for extracting and processing trajectory information from ROS bagfiles for motion evaluation.
- Temporal Stream Synchronization - Implements a time-series alignment mechanism to synchronize multiple trajectories by applying constant temporal offsets.
- ROS Bag Trajectory Extraction - Extracts trajectory data from ROS bagfiles including the ability to perform transform lookups between frames.
- Planar Pose Projections - Provides a visualization mechanism that flattens 3D spatial pose data onto specific 2D planes.
- Geographic Tile Overlays - Visualizes trajectories on real-world maps by overlaying geo-referenced tiles from external providers.
- Trajectory Pose Transformations - Transforms trajectory poses using custom 3D rotation and translation matrices applied from global or local frames.
- Experiment Result Comparators - Aggregates and contrasts performance metrics from multiple evaluation runs to compare trajectory accuracy.
- Temporal Trajectory Synchronization - Ensures temporal alignment between multiple trajectories by matching timestamps and applying constant offsets.
- Map Data Overlays - Inserts 2D ROS map images as custom data layers on top of trajectory plots for environmental context.
- Localización y estimación de estado - Python package for evaluating odometry and SLAM.
- Evaluation Tools - Command-line tools for evaluating trajectory and pose estimation accuracy.
