Utilizing Ego-centric Video to Conduct Naturalistic Bicycling Studies

Feng Liu, Portland State University



Bicycling is enjoying a surge in popularity across the USA but bicycle safety continues to remain a challenge. Cyclists (and pedestrians) are the most vulnerable road users since cyclist-vehicle collisions often result in severe injuries and have high fatality rates with respect to other types of collisions. In particular, intersections are one of the most critical areas in a road network given the high number of conflicts and accidents occurring at these locations. 

Existing data collection methods are mostly designed for videos captured by stationary cameras and are not designed to follow cyclists along a route or to integrate other sensor data. The goals of this research are: a) to develop a platform to collect naturalistic video bicycling data, b) to develop a methodology to integrate video data with other sensors that measure cyclists’ position and comfort levels, and c) to apply the platform and data collection methodology to a real-world route. This research effort has successfully integrated video and sensor data to describe cyclists’ comfort levels along a route. It was found that stress levels while riding during peak hours averaged 1.75 times higher than while riding at off-peak hours on the same routes and facilities. Separated bicycle infrastructure, such as multiuse paths, during peak and off-peak hours showed the lowest stress levels. Signalized intersections were hotspots for cyclists’ stress. All these results are statistically significant. The results indicate that integrating video and sensor data allows for a more detailed understanding of cyclists’ perceptions along a route. Rather than having an average measure for the whole route or path, it is possible to precisely identify the places and/or situations that trigger a change in experience or stress. By measuring how different facility types and riding conditions affect the distribution of stress levels among users, transportation engineers and planners may in the future incorporate video and detailed sensor data to evaluate the real-world performance of different types of facilities.

Project Details

Project Type:
Project Status:
End Date:
June 30,2016
UTC Grant Cycle:
Tier 1 Round 3
UTC Funding: