Thank you very much mister chairman, Ladies and Gentelmen. In this talk, I would like to share with you our initial experience with a new technology that provides drivers with feedback about their driving performance.
This technology is able to identify the occurrence of undesirable driving events? For example: Sudden braking or lane changing, sharp turning and excessive acceleration.
We call this technology the green-box. It consists of sensors for acceleration and speed and a processing unit able to identify driving events in real time . The data is transmitted to a server, which analyzes the data and generates a driver profile. A report generator then provides the driver with feedback via text messaging, e-mail or web-based reporting. Real time In-vehicle feedback is also available.
What you see here, ladies and gentlemen, is an example of the kind of information that this technology generates. Along the X axis we can see the days of the month, and along the Y axis we are see the number of trips taken each day, with each square representing a specific trip. The trips are color coded by safety level. A green trip has only few events, while a red trip means many undesirable events, or in other words, dangerous driving. Here we see a detailed description of the events that are included in one red square.
The study was conducted in Staffordshire county in the UK, were as in any other place in the world novice drivers are over presented in motor vehicle-related serious injuries and deaths. Thirty two drivers volunteered to participate in our study,
Our study design was based on comparing the driving performance of our volunteers between 2 periods. Initially, they drove with the green box activated, but they received no feedback. Then, during the feedback period, the drivers and their parents received feedback in real time while driving, by logging on to a web site, and via e-mail. Our volunteers accumulated a large volume of driving data for analysis, consisting of more than 18 thousand trips over 55 hundred hours. For the purpose of our analysis, the variable of interest was the event frequency which is the number of undesirable driving events per minute of driving.
For each of the 32 drivers, we calculated the event frequency. Here is a comparison between the blind profile period and the feedback period for each driver. As you can see, 28 drivers showed a reduction in undesirable events, and the great majority of these were statistically significant reductions.
Here is another look at the same results. this box plot gives us an idea about the distribution of the Event Frequency in the 2 phases. the mean frequency of undesirable events was reduced by MORE THAN HALF once feedback was enabled.
We continued our analysis by looking at other variables that may affect event frequency in new drivers, like the effect of timing. Using a mixed effect Poisson model, we estimated the event frequency by time of day (CLICK). and by day of the week (CLICK). we concluded that the effect of the feedback is consistent across times.
As for the effect of gender, both women and men show a similar reduction in their event frequency after receiving feedback. There was no significant difference between the sexes both in the blind profile period and in the feedback period.
One important question has been nagging us, was whether the reduction we see in the Events frequency is really the result of feedback, or perhaps it is merely the result of accumulated driving experience. To answer this question, we fitted several mixed effect Poisson models. A model that assumes neither experience nor feedback effect had the worse fit with our data. A model that assumes effect for experience only, fits a bit better. A model that assumes effect for feedback only fits even better. And a model that assumes effect for both feedback and experience has the best fit. Despite this It is important to note that the effect of experience is not statistically significant, but the effect of feedback is.
To FURTHER evaluate the model fit, we used the cumulative residual plot. WE plotted The cumulative residuals against time in the program, which is this case is the variable of interest. If the model fit is good then the cumulative residuals will oscillate around the zero line and within the control limits depicted by the red lines. so as you can see, in this case the model fit is quite good. This means that it is feedback, not experience, that plays the key role in improving the driving habits of novice drivers”.
Ladies and gentlemen, the take-home message of this presentation is that providing feedback from technology to young drivers about their driving habits can improve those habits and result in safer driving. This effect is consistent across times and across genders, and we were able to show that it is due to feedback, not accumulated experience, that the frequency of undesirable driving events decreases. We hope that this this pilot study will PAVE THE WAY for further evaluation of using technology to provide drivers with feedback.
On behalf of my co-author Liat Lampel and myself I would like to thank you very much for your attention and for the opportunity to present here today. Thank you very much!
When Technology Tells Novic Drivers How To Drive
Oren Musicant & Liat Lampel Ben - Gurion University, Israel When Technology Tells Novice Drivers How to Drive
What Are “Undesirable Events” ? Sudden braking Lane changing Sharp turning Acceleration
Server Reports generator Real-time feedback Sensors The “Green Box” Green Box
Square = 1 trip Color = safety level Web - Based Reporting
Staffordshire Driving Statistics <ul><li>Novice drivers in Staffordshire: </li></ul><ul><li>10% of population </li></ul><ul><li>32% of serious injuries / deaths </li></ul><ul><li>32 drivers volunteered </li></ul><ul><li>18 women, 14 men </li></ul>
The Study Design Event Frequency (EF) Blind profile period (3 months ) 18,240 trips over 5, 502 hours Green Box Feedback period (4 months )
Blind vs. Feedback Driving 28 showed EF reduction Significant (p<.05) in 25