HKUST-Aerial-Robotics/VINS-Fusion
VINS Fusion
VINS-Fusion is a multi-sensor fusion framework and visual-inertial odometry system. It integrates camera images, inertial measurement unit data, and global positioning signals through a non-linear optimization system to track the position and orientation of autonomous vehicles.
The system includes a visual loop closure engine that utilizes a bag-of-words approach to recognize previously visited locations and correct trajectory drift. It further provides tools for online spatio-temporal calibration to determine the physical offset and time synchronization between cameras and inertial sensors during operation.
Broad capabilities include multi-sensor state estimation and global coordinate alignment to ensure local trajectories are aligned with a world frame. The framework also supports camera model calibration to manage lens distortion and intrinsic parameters.
The project provides a containerized environment to ensure consistent build and execution processes across different host operating systems.
Features
- Sensor Fusion Frameworks - Implements a non-linear optimization framework that integrates visual, inertial, and GPS data for autonomous vehicle localization.
- Tightly-Coupled Fusion - Integrates raw camera imagery and IMU data into a single state vector for high-precision tracking.
- Loop Closure Detection - Includes a visual loop closure engine using a bag-of-words approach to recognize visited locations and correct trajectory drift.
- Online Sensor Calibrations - Performs real-time estimation of spatial offsets and temporal synchronization between cameras and IMUs.
- Sensor Fusion - Combines data from multiple sensors using non-linear optimization for autonomous vehicle localization.
- State Estimation Libraries - Fuses data from IMUs and cameras to track the robot's position and orientation in real time.
- Visual-Inertial Odometry Frameworks - Fuses camera imagery and IMU data to estimate the real-time position and orientation of a robot.
- Global Position Integration - Incorporates absolute global coordinates into local spatial estimations to align the robot trajectory with a world frame.
- Non-Linear Optimizers - Uses a sliding window of sensor measurements and iterative solvers to minimize estimation errors.
- Non-Linear State Estimation - Uses non-linear optimization of visual and inertial data to provide accurate self-localization.
- Camera Calibration - Calculates lens distortion and intrinsic parameters to ensure accurate visual measurements.
- Spatio-Temporal Calibrators - Provides tools for online spatio-temporal calibration to determine physical offsets and time synchronization between cameras and inertial sensors.
- State Marginalization - Optimizes computational efficiency by removing older states from the active window while preserving geometric constraints.
- सिमल्टेनियस लोकलाइज़ेशन और मैपिंग - Multi-sensor visual-inertial state estimator.
- State Estimation and SLAM - Optimization-based multi-sensor state estimation framework.
