REVIO: Range- and Event-Based Visual-Inertial Odometry for Bio-Inspired Sensors

Area of Science:

• Robotics and Computer Vision
• Sensor Fusion for Navigation

Background:

• Visual-inertial odometry (VIO) is crucial for Unmanned Aerial Vehicles (UAVs) and robotics but suffers from motion drift and blur in dynamic environments.
• Standard cameras struggle with sharp brightness changes and fast motion, degrading image quality and localization accuracy.
• Event cameras, inspired by biological vision, excel in high-dynamic-range scenes, offering a potential solution.

Purpose of the Study:

• To develop a novel range and event-based visual-inertial odometry (REVIO) system.
• To improve the accuracy and robustness of UAV localization in challenging visual conditions.
• To leverage the advantages of event cameras and range data for enhanced navigation.

Main Methods:

• Proposed an event-based visual-inertial odometry (EVIO) utilizing sliding window nonlinear optimization.
• Developed REVIO by integrating EVIO with range measurements to generate clearer event images.
• Incorporated additional range constraints into the optimization framework to refine position estimation.

Main Results:

• REVIO demonstrated reduced localization errors by nearly 29% compared to EVIO in handheld experiments.
• REVIO achieved almost 28% error reduction compared to VINS-Mono in flight experiments.
• The proposed method shows superior performance in fast-motion and high-dynamic scenes.

Conclusions:

• REVIO significantly enhances localization accuracy for UAVs in challenging dynamic environments.
• Fusing event camera data with range information is an effective strategy for robust VIO.
• The developed system offers a promising advancement for autonomous navigation in robotics and UAV applications.