The Clemson Vehicular Electronics Laboratory

Driver Arousal Monitor

sleepy driver

Most automotive systems today receive no data regarding the physiological or cognitive state of the user, but there are many cases where this data can be useful. For example, as the driver becomes bored or lethargic, systems can adjust to leave less room for driver error or provide audio-visual feedback to stimulate arousal. As the driver becomes tense or strained, an automobile's handling, ergonomics and displays can be adjusted in subtle ways to lessen arousal. The goal of this project was to build a monitoring device that produced a real-time cardiac-based measure of physiological arousal. The measure was based on changes in respiratory sinus arrhythmia, an established measure of vagal activity. The monitor provides computing systems with information about the physiological state of the user.


  • A. Hoover and E. Muth, A Real-Time Index of Vagal Activity, International Journal of Human-Computer Interaction, vol. 17, no. 2, 2004, pp. 197-209.
  • E. Muth, A. Kruse, A. Hoover and D. Schmorrow, "'Augmented Cognition: Aiding the Soldier in High and Low Workload Environments through Closed-Loop Human-Machine Interactions", in Military Life: The Psychology of Serving in Peace and Combat, edited by T. Britt, C. Castro and A. Adler, Praeger Security Int'l Publishing, 2006, pp. 108-127.
  • J. Rand, A. Hoover, S. Fishel, J. Moss, J. Pappas and E. Muth, Real-Time Correction of Heart Interbeat Intervals, in IEEE Trans. on Biomedical Engineering, vol. 54, 2007, pp. 946-950
  • S. Fishel, E. Muth, and A. Hoover, Establishing appropriate physiological baseline procedures for real-time physiological measurement, in Journal of Cognitive Engineering and Decision Making, vol. 1 no. 3, 2007, pp. 286-308.
  • S. Fishel, J. Owens, E. Muth, A. Hoover and J. Rand, "Augmented cognition: Developing and testing a physiology-based task adaptation system", in the proceedings of Human Factors and Ergonomics Society's 48th Annual Meeting, New Orleans, LA, October 2004.
  • L. Yu, A. Hoover and E. Muth, "Detection of Human Physiological State Change Using Fisher's Linear Discriminant", in the proc. of HCI International 2005.
  • E. Muth, A. Hoover and M. Loughry, "Developing an Augmented Cognition Sensor for the Operational Environment: The Wearable Arousal Meter", in the proc. of HCI International 2005.
  • J. Rand, A. Hoover, J. Pappas, J. Moss, S. Fishel and E. Muth, "Real-time correction of heart interbeat interval data", in the proc. of Biomonitoring for Physiological and Cognitive Performance during Military Operations, SPIE vol. 5797, edited by J. Caldwell and N. Wesensten, 2005, pp. 63-70.
  • S. Fishel, E. Muth, A. Hoover and L. Gugerty, "Determining the Resolution of a Real-Time Arousal Gauge", SPIE, vol. 6218, June 2006.


The latest version of the desktop arousal meter software is version 2.5b. Recent version upgrades made the following changes:

  • Version 2.3. Added Kalman filter for smoothing output.
  • Version 2.4. Added real-time IBI error detection and correction.
  • Version 2.5. Added various graphical displays (e.g. balloon) for non-numerical feedback.
  • Version 2.5b. Updated to connect to EZ-IBI2, which sends data synchronously instead of asynchronously.

We have created a tool ( IBIedit) for viewing and manual editing of IBI recordings.

The AViewer tool allows viewing and comparison of arousal files, along with simultaneous viewing of synchronous IBI traces.


In order to use our software, a heartbeat detector is required. We recommend using the latest version of the EZ-IBI, available from UFI, Inc. We have worked closely with UFI over the past few years to integrate our software closely with their hardware. They also sell a completely self-contained hardware arousal meter (the wearable arousal meter, or WAM), that embeds our software into their hardware. Contact UFI for more details.

Ongoing Efforts

How is a person's cognitive state related to his or her physiological state of arousal? Some preliminary evidence indicates they are related. We are in the process of a deeper study on this issue.

How can the information provided by the physiological monitor be used to close the human-computer loop? Our research suggests that changes in heart rate variability can be detected in periods as small as a few minutes. While this signal is not suitable for instant-action events, it shows promise for tracking human state over long-term activities.

What is the best way to analyze physiological data, to compute the physiological and cognitive "state" of the user? We are currently researching methods for subset Gaussian fitting of data in order to identify state.