Bargaining game approach for lane change maneuvers

Abstract

Lane change is one of the most common maneuvers in traffic and at least 60,000 people are injured from them. This problem is usually addressed from the perspective of automation and for a single vehicle using path planning and control approaches. However, traffic is a multi-agent system which can use communication technologies to perform coordinated maneuvers in a decentralized manner. This paper aims to present a Decentralized Bargaining Negotiation Process allowing that two conflicting agents negotiate. During the negotiation, agents calculate velocity change proposals and communicate with each other in order to solve the conflict at the lowest cost. This enables that agents open safe distance gaps and make the maneuvers. We also proposed a cost function attenuation so that the agents can continue to negotiate while the agreement was not reached, but with less impact on the cost. We simulated lane merge task on SUMO with a two-lane highway with speed limit of 30 m/s. We initially simulated 2 agents and we were increasing the number of agents up to 26. We recorded the total average time for the platoon to perform the lane merge task, the platoon average velocity and the average of the speed limit exceeded in each simulation. By increasing the number of agents the total average time stabilized approximately 25 seconds. Which represents an expressive result because the time for one vehicle to complete the maneuver is 4.6 seconds. The average speed was also approximately constant around 22 m/s, which represents a good result for the traffic flow. The highest speed exceeded 9 % which represents 2.70 m/s with the attenuation cost function. We can observe that the approach proved to be scalable since the analyzed data show stability by increasing the number of vehicles.