Examining the driverless future: An analysis of human-caused vehicle accidents and development of an autonomous vehicle communication testbed

This design study examines the trade-offs that occur in autonomous vehicle hardware and software design. Through this research process, the creation of a research testbed allows for more in-depth and holistic understanding of the design and construction process of scaled autonomous vehicles. The two current industry approaches to autonomous driving are the “Google Car Paradigm,” where vehicles rely on sensing of the physical world to act accordingly, and the “V2X Paradigm,” which additionally enables inter-vehicle communication. The latter approach has the potential to provide increased safety, mobility, and environmental sustainability, however, these claims have limited verification to date. Construction of a scaled autonomous vehicle testbed provides the necessary foundation for future teams to analyze the trade-offs between the two communication protocol in a variety of experimental designs to test environmental impact and traffic flow, among other research questions. Analysis of historical Virginia crash data was conducted to determine the conditions under which automation will reduce crash severity. Historical data from the years of 2011 to 2017 were utilized and segmented into categories inclusive of geographical region and details surrounding the accident such as the number of casualties, collision type, distractions, etc. Results yielded that crashes caused by human-error were found to be higher in severity, as well as a higher number of casualties, when alcohol was involved and when drivers self-reported drowsiness during the accident, both of which are crashes caused by human factors. These are situations in which autonomous vehicles would be able to minimize the likelihood of a crash, reduce the overall crash severity, and decrease human casualties. The design of the scaled vehicle testbed was designed with the capability to properly test metrics in areas of environmental impact, traffic flow, crash safety, and cyber-security to properly evaluate and compare the effectiveness of the two communication paradigms, as well as other metrics future experimenters develop. The vehicle design has the ability to answer important communication questions without the need for expensive equipment and substantial human risk. Additionally, this framework can be easily duplicated by other organizations and educational institutions to increase the collective knowledge of communication protocol and increase intercollegiate research.