Driverless Planning and Decision-Making based on GRU

Abstract

In the wake of the swift advancement of artificial intelligence, driverless technology has progressively emerged as a revolutionary technology. The existing driverless technology primarily relies on advanced sensors, artificial intelligence and machine learning algorithms, with a view to perceiving and analyzing the surrounding environment in real time, and making driving decisions accordingly. At the present time, despite the fact that driverless technology has introduced a lot of convenience to human travel, it still encompasses a significant number of challenges, such as safety, perception technology, vehicle control, road planning, cost and other issues. In a similar vein, the safety issue is an unavoidable part of the discussion on the driverless. Driverless vehicles require robust system safety and protection mechanisms. This paper is designed to establish a data model by analyzing the NGSIM dataset, and to probe the driverless perception capability and decision-making on intelligence in behavior-aware motion perspective with GRU approach. Eventually, it was concluded that neither too many nor too few neighboring vehicles should be taken into account when forecasting the trajectory of each vehicle in the coming years, while the appropriate neighboring vehicles should be selected to achieve the best prediction results.