Donald K. Grimm

Research in vehicular analytics has explored two different approaches to infer properties about a vehicle's surroundings: (a) using sensors on smartphone devices to infer properties about surroundings, or (b) to use in-vehicle sensors. The latter approach was less studied. In this paper, we take a first step beyond research to understand, using a pilot study, how to design vehicular analytics at scale. Our pilot prototype NORA (Network Oriented Road Applications) contains novel algorithms to detect roadside phenomena (such as potholes, rough road, and slippery surfaces). These leverage multiple in-vehicle sensors to disambiguate these phenomena from other conflating factors. NORA reliably detects, on a cloud service, roadside events from multiple individual in-vehicle detections. To do this, it uses careful clustering techniques to assess the spatial scale of the event, and belief decay techniques to match event duration. It also employs aggressive fleetwide suppression of detections to minimize communication cost. Through a 50-vehicle deployment of NORA pilot over 9 months, it is shown that the results obtained from NORA in-vehicle detection methods match very well with ground truth measurements, and NORA cloud is effective at aggregating road events in an accurate and efficient fashion.

This work presents an approach for collecting road surface data using connected vehicles. Road surface readings from multiple production vehicles were collected and aggregated to estimate road roughness measured by the International Roughness Index (IRI). The analysis compared multiple instances of connected vehicle data with high speed pavement profile vehicle (Class 1 profiler) data. A separate analysis compared multiple instances of connected vehicle data to an advanced walking profiler. Results demonstrate the feasibility of harvesting road surface data from the existing connected vehicles to support continuous road surface monitoring applications. Benefits include more timely acquisition of pavement data, broader coverage of the road network, and potential for aiding existing survey fleet in targeting early signs of pavement degradation. Collected roughness measurements were found to be closely aligned with reference devices that were employed as part of this study. A regional experiment in the Detroit Metropolitan area that covered 64 mi of roadways found that the connected vehicle data was highly correlated with Class 1 profiler data where 83% of traveled miles had a 0.8 or higher correlation. Moreover, 85% of the measurements had small absolute errors less than 50 in./mi and half of the measurements had absolute errors less than 20 in./mi. A test track experiment at Virginia Tech Transportation Institute Smart Road facility compared the connected vehicle data to the advanced walking profiler and showed that the correlations for repeatability and reproducibility are 0.90 and 0.91, respectively, which are very close to the standard requirement for certified profilers.

In-vehicle context sensing can detect many aspects of driver behavior and the environment, such as drivers changing lanes, stop signs, road grade and potholes, and these features can be used to improve driver safety and comfort, and engine efficiency. In general, detecting these features can use either on-board sensors on the vehicle (car sensors), or sensors built into mobile devices (phone sensors) carried by one or more occupants, or both. Furthermore, traces of sensor read- ings from different cars, when crowd-sourced, can provide increased spatial coverage as well as disambiguation. In this paper, we explore, by designing novel detection algorithms for the three different features discussed above, three related questions: How is the accuracy of detection related to the choice of phone vs. car sensors? To what extent, and in what ways, does crowd-sourcing contribute to detec- tion accuracy? How is accuracy affected by phone position? We have collected hundreds of miles of vehicle traces with annotated groundtruth, and demonstrated through evaluation that our detection algorithms can achieve high accuracy for each task (e.g. > 90% for lane change determinations) and that crowd-sensing plays an indispensable role in improving the detection performance (e.g. improving recall by 35% for lane change determinations on curves). Our results can give car manufacturers insight into how to augment their internal sensing capabilities with phone sensors, or give mobile app developers insight into what car sensors to use in order to complement mobile device sensing capabilities.