Pedestrian Dead Reckoning-Assisted Visual Inertial Odometry Integrity Monitoring

Area of Science:

• Robotics and Computer Vision
• Indoor Navigation Systems
• Sensor Fusion

Background:

• Visual inertial odometers (VIOs) are crucial for indoor positioning, leveraging camera convenience.
• VIO accuracy is compromised by environmental factors like low illumination and lack of texture.
• Existing VIO systems are susceptible to visual observation errors, impacting precise localization.

Purpose of the Study:

• To analyze visual observation errors in VIO under various environmental conditions.
• To propose a novel method for enhancing VIO reliability using pedestrian dead reckoning (PDR).
• To reduce indoor positioning error by integrating VIO and PDR.

Main Methods:

• Investigated causes and impacts of visual observation errors in different scenarios.
• Developed a PDR-assisted visual integrity monitoring system.
• Implemented an automatic switching mechanism between VIO and PDR outputs for optimized positioning.

Main Results:

• The proposed system demonstrated improved performance in indoor positioning tasks.
• Effectiveness was validated in environments with low illumination and scarce texture.
• The system successfully reduced positioning errors compared to VIO alone.

Conclusions:

• PDR-assisted visual integrity monitoring significantly enhances VIO robustness.
• The proposed method offers a reliable solution for accurate indoor positioning.
• Automatic switching between VIO and PDR provides adaptive and precise localization.