Warwick Logo JLR Logo

Home

This page is for accessing the Warwick-JLR Driver Monitoring Dataset (DMD) and the Road Classification Dataset (RCD).


If you have not done so already, please email Phillip.Taylor@warwick.ac.uk with your research interests and outline why the DMD may be helpful to you. We are interested in driver monitoring research and data science generally, and would like to hear how the DMD is being used in the community. Further, it enables us to update you on our progress if you are interested.


The Warwick-JLR Driver Monitoring Dataset

The DMD is detailed in our Applied Artificial Intelligence paper, "Investigating the Feasibility of Vehicle Telemetry Data as a Means of Predicting Driver Workload". It was collected as part of the Driver Monitoring: Assessing Cognitive Load project, funded by EPSRC and Jaguar Land Rover (2011-2015). The DMD was collected and analysed as part of work for the PhD thesis, "Data Mining for Vehicle Telemtry Data" (Phillip Taylor, 2015).

The DMD can be downloaded using the following links:

In both data streams samples are listed in rows, the CAN data is recorded at 20hz and the BIO data is recorded at 256hz.

JLR requested that the signals in the CAN-data remain nameless. If you are interested in a particular signal, please email and enquire. The first 8 columns of the CAN-data (after resampling and normalisation in the range [0,1]) are as follows:

  1. Label_ACTION
  2. GTEC_TIME
  3. GT_HR
  4. GT_HRV
  5. ECG
  6. GT_EDR
  7. GT_SC
  8. EDA

The signals columns in the Biosignals data are 'Label_ACTION','ECG','GT_HR','GT_HRV','EDA','GT_EDR','GT_SC','GTEC_TIME'.

Label_ACTION shows what the driver's were doing when, listed in numbers 1-16:

  1. Video Start
  2. Start moving
  3. Join track
  4. Baseline
  5. 0 back intro
  6. 0 back
  7. 0 back recovery
  8. 1 back intro
  9. 1 back
  10. 1 back recovery
  11. 2 back intro
  12. 2 back
  13. 2 back recovery
  14. Leave track
  15. Car stop
  16. Video stop

Recommended citations for the Driver Monitoring Dataset are as follows:

Phillip Taylor, Nathan Griffiths, Abhir Bhalerao, Xu Zhou, Adam Gelencser, Thomas Popham. 2017. Investigating the feasibility of vehicle telemetry data as a means of predicting driver workload. International Journal of Mobile Human-Computer Interaction, 9(3).

Phillip Taylor, Nathan Griffiths, Abhir Bhalerao, Xu Zhou, Adam Gelencser, Thomas Popham. 2015. Warwick-JLR Driver Monitoring Dataset (DMD): Statistics and Early Findings. In the 7th International Conference on Automotive User Interfaces and Interactive Vehicular Applications.

Phillip Taylor, Nathan Griffiths, Abhir Bhalerao, Derrick Watson, Xu Zhou and Thomas Popham. 2013. Warwick-JLR Driver Monitoring Dataset (DMD): A public Dataset for Driver Monitoring Research. In CLW 2013: The Third Workshop on Cognitive Load and In-Vehicle Human-Machine Interaction.


The Road Classification Dataset

The RCD can be downloaded here, with a sample rate of 20Hz.

Recommended citations for the Road Classificication Dataset are as follows:

Phillip Taylor, Nathan Griffiths, Abhir Bhalerao, Sarabjot Anand, Thomas Popham, Zhou Xu, Adam Gelencser. 2016. Data Mining for Vehicle Telemetry. In Applied Artificial Intelligence 30(3): 233-256.

Phillip Taylor, Sarabjot Anand, Nathan Griffiths, Fatimah Adamu-Fika, Alain Dunoyer, Thomas Popham. 2012. Road Type Classification Through Data Mining. In the 4th International Conference on Automotive User Interfaces and Interactive Vehicular Applications.