Inertial Sensors - Applications and Challenges in a Nutshell

Thomas Seel, Manon Kok, Ryan S. McGinnisSensors 2020 · OpenReview

TL;DR

This editorial provides a concise overview of inertial sensors, highlighting their widespread applications and the significant signal processing challenges they present. It serves as an introduction to a special issue, guiding readers through the current state of research in inertial sensor methods and applications, from robotics to healthcare.

What Problem Does This Paper Address?

Inertial sensors, such as accelerometers and gyroscopes, are becoming ubiquitous in modern technology, from smartphones to intelligent vehicles. They are crucial for tasks like localization, navigation, and mapping, as well as for capturing human movement in health and sports applications. However, despite their miniaturization and cost reduction, inertial sensors present substantial signal processing challenges. These include:

  • Coordinate System Mismatch: Sensor measurements are in the sensor's moving frame, requiring complex integration to relate to the real world. Aligning sensor frames with object frames, especially with flexible or soft tissues, is also difficult.
  • Magnetic Disturbances: In indoor environments, magnetic fields can be inhomogeneous, making accurate heading estimation unreliable and complicating orientation estimation.
  • Signal Corruption: Inertial sensor signals are prone to time-variant biases and noise, leading to integration drift and erroneous position/orientation estimates.

This paper addresses the need for a clear understanding of these challenges and the diverse methods and applications that have emerged to overcome them, setting the stage for the research presented in the special issue.

Methods

This editorial doesn't present novel methods but rather summarizes existing and emerging approaches used with inertial sensors. Key methods discussed include:

  • Strapdown Integration: A fundamental technique for integrating inertial sensor data to estimate orientation and position.
  • Sensor Fusion: Combining data from inertial sensors with other sources (e.g., additional sensors, mathematical models) to improve accuracy and mitigate drift. This includes various filtering approaches like Extended and Unscented Kalman Filters, and complementary filters for real-time estimation, as well as offline smoothing techniques.
  • Kinematic Models: Incorporating biomechanical or mechanical models to aid estimation, particularly in human motion tracking, reducing reliance on additional measurement modalities.
  • Machine Learning: Increasingly used for classification (e.g., human activities) and regression (e.g., estimating walking speed, learning sensor orientation, vehicle motion estimation, localization, and mapping).
  • Sensor-to-Segment Calibration: Methods for accurately aligning sensor frames with body segments, especially important in biomechanics where attachments might not be rigid.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • Ubiquity and Growing Importance: Inertial sensors are no longer niche components; they are fundamental to a vast array of modern applications, driven by hardware advancements.
  • The "Nutshell" of Challenges: The core difficulties with inertial sensors boil down to integrating data from a moving frame, dealing with environmental interference (especially magnetic), and correcting for inherent sensor errors.
  • Fusion is Key: Due to the inherent limitations of individual inertial sensors, sophisticated sensor fusion techniques are paramount for achieving reliable motion estimation.
  • Machine Learning's Expanding Role: ML is proving to be a powerful tool for extracting meaningful information from inertial sensor data, enabling complex tasks like activity recognition and precise motion estimation.
  • Bridging the Gap: The special issue aims to showcase how research is actively addressing these challenges across diverse fields, from robotics to healthcare.

Key Results

As an editorial, this paper does not present quantitative results. Instead, it highlights the breadth of applications and the types of challenges addressed by the papers in the special issue. These include:

  • Robotics: Papers focus on orientation estimation for mobile robots, stabilizing optical payloads in UAVs, and trajectory estimation for ground robots.
  • Human Biomechanics, Gestures, and Activities: A significant portion of the special issue is dedicated to wearable inertial sensors for capturing human movement. This includes new approaches for estimating joint angles (knee, hip) and forces, as well as systematic reviews of lower limb joint kinematics.
  • Health and Sports Engineering: Applications range from detecting disease and risk status to estimating gait speed and analyzing hand gestures.

Limitations and Future Work

The editorial acknowledges the ongoing challenges in inertial sensing, particularly concerning:

  • Accurate Calibration: Especially in biomechanics, achieving precise sensor-to-segment calibration remains an area of active research.
  • Robustness to Magnetic Disturbances: Developing methods that are less susceptible to magnetic interference is crucial for reliable indoor navigation and orientation estimation.
  • Minimizing Integration Drift: Continued development of advanced fusion and estimation techniques is needed to combat the inherent drift in inertial measurements.

The paper suggests that future work will likely continue to focus on solving these challenges, further expanding the reach and establishment of inertial sensor technologies across a wide range of application domains.

Summary updated

Abstract

This editorial provides a concise introduction to the methods and applications of inertial sensors. We briefly describe the main characteristics of inertial sensors and highlight the broad range of applications as well as the methodological challenges. Finally, for the reader’s guidance, we give a succinct overview of the papers included in this special issue.
BibTeX
@inproceedings{seel2020inertial,
  author = {Thomas Seel and Manon Kok and Ryan S. McGinnis},
  title = {{Inertial Sensors - Applications and Challenges in a Nutshell}},
  booktitle = {Sensors},
  year = {2020},
  url = {https://openreview.net/forum?id=S2PY1d8oDr1},
}
Figure