Analysis of the deployment of autonomous vehicles, in high pedestrian demand areas in Malta, through micro-simulation tools

Abstract

Autonomous vehicles (AVs) are self-driving vehicles which are foreseen to aid in improving transportation systems in relation to traffic performance by increasing road safety and efficiency. The interaction between such technology and other road users, such as pedestrians, has proved to be complex, and is currently a major concern which requires further study prior to full deployment on intricate road networks. Many countries worldwide are researching and testing the autonomous notions and their availability in future years. However, testing on real road networks can be very challenging due to strict regulations and legislation, and thus, simulating this technology in a virtual world can help decisionmakers to assess the possible benefits and risks, and predict potential challenges. Using AIMSUN Next 22 micro-simulation software, AVs and pedestrians were simulated on a road network to better understand their interaction. Multiple scenarios with different AV penetration rates, pedestrian demand levels, and frequency of zebra crossings in a mixed-use traffic environment were tested for both a fictitious and a real-world case study location. The fictitious case study road network that was simulated consists of a single-lane straight road having a speed limit of 35 km/hr and being 2 km long. On the other hand, the real-world case study road network was chosen to be the landside of the Malta International Airport (MIA), consisting of a single-lane ring road having a speed limit of 35 km/hr and being 1km long. From the results obtained, it was found that when subjected to higher levels of pedestrian penetration rates, AVs are negatively affected with respect to their traffic performance indicators such as average speed, travel time and delay time. The results indicated that with higher pedestrian penetration rates at crossing situations, the difference in results achieved between the conventional Human-Driven vehicles (HVs) and AVs began to diminish. In fact, AVs started to perform as good, and in some cases better, than HVs when deployed in areas of high pedestrian demand. Since HVs have the potential to drive at faster speeds, their average speed was reduced more than that of AVs, whilst their average travel time and delay time were increased. The parameters which allowed AVs to perform better than HVs were the faster reaction times and the lower minimum gaps between vehicles. Moreover, through an analysis of the average pedestrian speed in different levels of AV penetration rate, it was found that pedestrians moved faster as the AV penetration rate increased, implying that pedestrians were more hesitant to cross the road with higher percentages of HVs present within the road network. Therefore, the findings of this study indicate that, in addition to the theoretically safer interaction between pedestrians and AVs, they are also more effective on the road network when compared to HVs. It was concluded that in a future scenario where AVs drive at similar speeds to HVs, the deployment of AVs within the transportation network will see an improvement in traffic flows, more specifically when incorporated in high pedestrian demand areas. Also, through the deployment of Shared Autonomous Vehicle (SAV) services, not only would these benefits be utilised, but their adoption would also reduce the travel demand, which would further improve the road traffic performance.