A study at Tsinghua University in Beijing investigated the dynamics of a road that contains both automated vehicles and human-operated vehicles. This scenario indicates the intermediate state between contemporary behavior and a future conversion to a road of entirely automated vehicles. The heterogenous mix of automated and human-operated vehicles is of interest because many of the benefits of completely automated fleets, such as being able to handle significantly closer following distances, are mitigated by the presence of human operators. However, the transitional phase is expected to last a substantial amount of time, and thus it is important to understand the dynamics of how human-operated vehicles interact with fully automated vehicles.
The primary subject of this investigation was the comparison of four common rear-end collision avoidance algorithms, and a simulation was carried out to examine how these algorithms performed under various automated-vehicle market penetration rates. The study used a simulation platform designed and constructed in MATLAB, which allowed for more flexible data processing. Eleven scenarios were examined, with automated-vehicle market penetration rates ranging from 0 to 100 percent. The average rates of crashes for each car an 11-car fleet were examined and used to evaluate each collision avoidance algorithm's performance.
- The Safe Distance algorithm was also found to have the strongest efficacy in eliminating crashes at market penetration rates below 80 percent. At 40 percent market penetration, the algorithm resulted in less than half as many predicted collisions.
- The Total Relative Kinetic Energy Density algorithm was found to have the potential to eliminate crashes altogether at a penetration rate of 100 percent, while also being the most effective at minimizing the severity (but not occurrence) of crashes at any market penetration rate.