HeadSLAM: Pedestrian SLAM with Head-Mounted Sensors.

Area of Science:

• Robotics and Human-Computer Interaction
• Sensor Fusion and Navigation

Background:

• Human position tracking with wearable sensors is crucial for applications in healthcare, smart homes, sports, and emergency services.
• Pedestrian Dead Reckoning (PDR) using Inertial Measurement Units (IMUs) offers infrastructureless navigation but suffers from drift-induced errors over time.
• Accurate long-term position estimation is limited by PDR's inherent inaccuracies and calibration challenges.

Purpose of the Study:

• To introduce and evaluate HeadSLAM, a novel algorithm for enhanced accuracy in human position tracking using head-mounted IMUs.
• To compare the performance of HeadSLAM against traditional PDR methods for long-term tracking.
• To demonstrate the potential of HeadSLAM for reliable, low-cost, infrastructureless navigation solutions.

Main Methods:

• Developed and implemented the HeadSLAM algorithm specifically for head-mounted IMUs.
• Recruited 7 research participants to walk in diverse indoor and outdoor environments while wearing head-mounted sensors.
• Compared the accuracy of HeadSLAM against PDR using metrics such as average root-mean-squared error and absolute error over a 20-hour walking dataset.

Main Results:

• HeadSLAM demonstrated significantly lower average root-mean-squared error and absolute error compared to PDR (p < 0.001).
• Consistent error reduction was observed across all participants and scenarios in the extensive 20-hour dataset.
• HeadSLAM proved more accurate than PDR for long-term position tracking, mitigating drift-related inaccuracies.

Conclusions:

• The HeadSLAM algorithm significantly enhances the accuracy of long-term human position tracking using low-cost, head-mounted sensors.
• HeadSLAM offers a viable solution for infrastructureless navigation, overcoming the limitations of traditional PDR.
• This advancement supports the development of more reliable and affordable navigation applications.