AI Algorithms for Autonomous Vehicle Decision-Making in Complex Traffic Scenarios

Abstract

The performance of the trained models was evaluated in unknown environments, in this case a faculty connected to the testing site with a communitarian road. In a known environment, the system was capable of executing navigation passing through critical situations. Performance evaluation was relational to a set of indexes regarding requirements, for example, collision avoidance, detection of persons, and detection of static and dynamic obstacles or limitations. The two manned vehicles for validating the baseline data were the driverless vehicle communicating with the call center of the faculty and the validation of the navigation route using the GPS codes of this system. All these new would like to respond to the following question: What are the probable problems, methods, solutions, and impacts regarding the extensive use of driverless vehicles in urban traffic and some rural and highway situations? Major attention was paid to the different types of driverless vehicles that were used in real traffic that showed that there is a mismatch between the recent efforts of the research team and the real-world experiences of commuters (on the drivers’ and partners’ side) that moped, motorcycles, push scooter, or electric bike riders have been affected by fear and unease.