Hybrid Indoor Localization Using IMU Sensors and Smartphone Camera

Area of Science:

• Robotics and Computer Vision
• Sensor Fusion for Localization

Background:

• Smartphone cameras and Inertial Measurement Unit (IMU) sensors offer independent indoor positioning capabilities.
• IMU-based systems face accuracy issues due to sensor errors like electromagnetic noise and drifts.
• Camera-based systems are limited by reliance on floor maps, camera poses, and rapid directional changes.

Purpose of the Study:

• To develop and evaluate hybrid indoor localization systems that integrate smartphone camera and IMU sensor data.
• To minimize positioning errors inherent in standalone camera or IMU localization methods.
• To analyze the performance of hybrid systems compared to individual approaches.

Main Methods:

• Designed an indoor experiment to assess IMU-based, camera-based, and hybrid localization systems.
• Implemented hybrid systems by fusing camera-based techniques (ORB-SLAM, UcoSLAM) with IMU sensor data.
• Evaluated system performance using metrics like average localization error, mean error, maximum error, and standard deviation.

Main Results:

• The proposed hybrid systems demonstrated superior position accuracy compared to individual IMU and camera-based systems.
• The ORB-SLAM with IMU hybrid system achieved a mean localization error of 0.1398 m.
• The UcoSLAM with IMU hybrid system achieved a lower mean localization error of 0.0690 m.

Conclusions:

• Hybrid systems combining camera and IMU sensors are effective for enhancing indoor localization accuracy.
• The UcoSLAM with IMU hybrid system showed particularly strong performance in the conducted experiments.
• Sensor fusion offers a robust solution to mitigate individual sensor limitations in indoor positioning.