Webinar Question and Answer Transcript

Britton Hammit: When looking at acceleration data we’re using this relative difference in acceleration and that’s when we conducted the first four trips in sunny or what would be considered normal or good weather conditions so that then we could compare what is the normal acceleration rate in the sunny conditions compared with bad weather conditions—adverse weather conditions. That’s another reason why we did the sunny and non-sunny adverse weather. That’s also why we use the exact same quartier for each of the test trips so that when there was some type of surprising acceleration or deceleration, we can look and compare the windshield data and possibly see why there was a dramatic difference in acceleration. Acceleration is a tricky parameter. There are a lot of reasons for different acceleration rates. For example, you’re in the car with your kids and your kids are acting out and you pull to the side of the road. In any of these vehicle parameters analysis, we have to combine it with a lot of other data. That’s why we have to aggregate this weather data and aggregate other vehicle internal data so that we are not just looking into acceleration and trying to decide that it must be adverse weather-related.

Dr. Rhonda Younge: I thought maybe you could just talk about your experiences dealing with the amounts of data—the frequency of the data that was transmitted. Can you comment on what the experience with big data analysis was like?

Britton Hammit: As mentioned, we had 50 million data from just 16 test trips. These were just 16 80-mile test trips. Considering this concept, but on an even larger scale, it has to do with a lot of data aggregation. So myself and a computer science student helped me. We had to break down the data. Basically, we took averages for some parameters. For windshield wiper data, we took readings for every two seconds. For acceleration, we took 3-5 aggregations per second so that it became more processable and more comprehendible.

Dr. Rhonda Younge: We are currently in the process. I cannot talk about specifics. I would say that probably another month that would be considered more public information. In terms of what specific data—certainly the temperature, the air temperature, dew points are important ones. I’ve done a lot of modeling for the variable speed limit system—the temperature and relative humidity are pretty significant variables for that. I’m not sure that anything with the road friction would be coming from the road weather data. We are fortunate in that the I-80 corridor has a pretty good instrumentation just given the weather hazard the corridor has and the variable speed limit. We will use the weather data to look for the gaps and changes in the system. We will have better answers within the next month.

Britton Hammit: ABS braking and the traction stability control are really important for this type of application. Because they are pretty much one of the only parameters any vehicle that you don’t really see in other conditions. And that’s not entirely true because if you are driving with traction stability activations and there are other reasons for ABS brakes. But I think that for DOTs to start asking to Operations & Information Management (OIM) to standardize for researchers to gather that information for the vehicles would make this research meaningful.

Britton Hammit: Again, if you email me directly, at the email address provided on the PowerPoint, I would be happy to provide you with some information from my research.

Dr. Rhonda Younge: I know that in the slides there is also a link to the full research report which is available through the mountain consortium, which would have a full list of variables.

Britton Hammit: In the scope of our projects, we did not have vehicle video data. I think that’s the next step. Right now, the University of Wyoming is involved in a project that has data that we are trying to evaluate—this video information data. That would be the next step—combining visual analysis with this pure data internal vehicle and possibly external sensors.

Dr. Rhonda Younge: So that the data set that she was referring was the National Safe Driving Study—looking specifically at weather events and so that has both the vehicles data, forward-looking vehicles, and in-cab video.

Dr. Rhonda Younge: They have surface cuts that they use in the pavement layer and it leads to surface condition rating—not specifically on friction, but surface condition whether it is icy, snowy, or chemically, which could be converted into ranges of friction.