An Evaluation of MEMS-IMU Performance on the Absolute Trajectory Error of Visual-Inertial Navigation System

Area of Science:

• Robotics and Autonomous Systems
• Sensor Fusion
• Navigation Systems

Background:

• Accurate robot localization is crucial for autonomous systems.
• Sensor fusion, particularly visual-inertial odometry (VIO), enhances localization accuracy.
• The impact of varying IMU performance on VIO accuracy is under-explored.

Purpose of the Study:

• To investigate the influence of different IMU grades on visual-inertial system (VINS) accuracy.
• To provide guidance on selecting appropriate IMUs for large-scale hardware deployments.
• To analyze the trade-offs between IMU performance and VIO results in diverse scenarios.

Main Methods:

• Evaluated six representative IMUs ranging from consumer-grade to tactical-grade.
• Performed VIO using the VINS_Fusion framework with each IMU.
• Analyzed absolute trajectory error across eight distinct experimental scenarios.

Main Results:

• IMU assistance generally improves multi-sensor fusion accuracy.
• The accuracy improvement from different MEMS-IMU grades was not consistently significant across scenarios.
• Consumer-grade IMUs achieved excellent results in many cases.
• Low-noise IMUs demonstrated greater versatility and stability.

Conclusions:

• The performance difference between various IMU grades may not be substantial for VIO accuracy in many applications.
• Consumer-grade IMUs are a viable option for VIO, offering cost-effective solutions.
• Selecting IMUs with lower noise is recommended for enhanced stability and versatility in diverse operational environments.
• This research provides a practical guideline for IMU selection in VIO and Inertial Navigation System (INS) development.