Autonomous vehicles' performance on single lane road: A simulation under VISSIM environment

This paper tests an autonomous vehicles (AV) longitudinal driving model Under VISSIM environment. The AV will obey three main rules: “Assuring a Clear Distance Ahead”, “Keeping Away from Dilemma Zone”, and “No Speeding”. To evaluate the performance of the AV, a simulation environment based on VISSIM was constructed. According to the simulation experiments, the average capacity per lane for AV in a single-lane road segment is about 2233 vehicles/hour, which is 1.13% higher than that for Human-Driving Vehicle (HV). However, AV will have a much better performance under heavy traffic condition because of its less desired speed deviation. As for signalized entrance lane, it shows that AVs prefer a more conservative driving way, which lead to a quick reduce in number of vehicles passing stop line during the beginning of amber period, which means lower possibilities for traffic conflict during the signal switching time. However, the total flow rate for AV during the whole signal time is almost the same as HV because AV has a higher flow rate during the middle and end of green time to make up the loss in amber time.