Self-driving cars improve traffic — even when most other vehicles are driven by people

Robotic vehicles have shown promising potential in optimizing traffic flow within cities, even when operating alongside human-driven vehicles. This advancement aims to enhance traffic efficiency, safety, and energy consumption, as observed by a group of researchers.

The utilization of autonomous robotaxis in cities globally since 2016 has marked a turning point, transitioning robotic vehicles from a concept in science fiction to a reality.

Considering the increasing integration of robot vehicles into traffic and the prolonged transition period towards fully autonomous traffic, a team of researchers delved into exploring the potential of robot vehicles to alleviate prevalent traffic issues when interacting with human-driven vehicles.

With a focus on artificial intelligence for transportation and smart cities, the researchers hypothesized that leveraging AI algorithms could effectively manage the complexities of mixed traffic systems as the number of robot vehicles on the roads grows.

Through the application of reinforcement learning, a branch of AI enabling intelligent agents to maximize rewards through interactions, the researchers developed algorithms to optimize traffic flow and coordination among robot and human-driven vehicles.

Notably, their experiments demonstrated that by setting specific goals for simulated robot vehicles, such as prioritizing traffic efficiency or energy consumption, the mixed traffic system at real-world intersections could be effectively managed under diverse traffic conditions in simulations.

Their algorithm empowered the robot vehicles to enhance traffic flow by coordinating with each other, enabling smooth traffic progression as each vehicle made intersection decisions based on its surroundings.

Remarkably, the research findings indicated that introducing just 5% of robot vehicles into traffic resulted in the elimination of traffic congestion, with traffic efficiency surpassing that of traffic lights when robot vehicles constituted 60% of the traffic.

The implications of this research are significant, as cities worldwide grapple with worsening traffic conditions that pose economic and environmental challenges. Traditional traffic control methods have shown limited effectiveness in reducing congestion and delays, underscoring the potential of AI-driven robot vehicles as a solution.

While previous studies often assumed universal connectivity and centralized control of all robot vehicles, a scenario unlikely to occur promptly, this research focuses on developing control algorithms that leverage the benefits of autonomous transportation systems without mandating all vehicles to be autonomous.

Noteworthy advancements have been made in mixed traffic control studies across various scenarios, but this research stands out as the first to showcase the feasibility of controlling mixed traffic at complex real-world intersections using robot vehicles.

Future plans for the research include expanding the framework to incorporate additional driving behaviors for robot vehicles and testing the approach across diverse intersection types, aiming to achieve effective and efficient mixed traffic control on a city scale.

The Research Brief is a short take on interesting academic work.

Weizi Li, Assistant Professor of Computer Science, University of Tennessee

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Source: MyBroadband