Driverless vehicle motion trail planning system and method for structured road

Abstract

The invention relates to a driverless car motion trail planning system for a structured road. The driverless car motion trail planning system comprises a sensing module, a positioning module, a lane changing decision module, a motion planning module and a trajectory tracking module. The lane changing decision module outputs a decision action according to the data acquired by the sensing module andthe positioning module; and the motion planning module outputs an optimal trajectory to the trajectory tracking module according to the decision action. Deep reinforcement learning is used for makinga decision, a trajectory planning module, a perception module, a control module and other modules based on decision action dynamics are independently processed, the interpretability and operability of the decision planning process are greatly improved compared with an end-to-end method, and the invention can be well adapted to a previous driverless automobile system architecture.

Description

technical field [0001] The invention belongs to the technical field of unmanned driving, and in particular relates to a motion trajectory planning system and method for unmanned vehicles aimed at structured roads. Background technique [0002] The goal of the decision-making and motion planning system of unmanned vehicles is to make unmanned vehicles produce safe and reasonable driving behavior like a skilled driver, and plan a safe driving trajectory. Traditional decision-making planning systems are mostly rule-based decision-making planning methods, which have clear logic and strong planning and reasoning capabilities, but they need to anticipate various scenarios that unmanned vehicles will encounter, and the system is relatively complicated. In recent years, the method of decision-making planning using end-to-end convolutional neural network has been applied to unmanned vehicles, which greatly simplifies the decision-making planning system. The system directly inputs eac...

AI Technical Summary

Problems solved by technology

Traditional decision-making planning systems are mostly rule-based decision-making planning methods, which have clear logic and strong planning and reasoning capabilities, but they need to anticipate various scenarios that unmanned vehicles will encounter, and the system is relatively complex

In recent years, the method of decision-making planning using end-to-end convolutional neural network has been applied to unmanned vehicles, which greatly simplifies the decision-making planning system. The system directly inputs each frame of image obtained by the camera, and directly outputs the target steering of the vehicle after making a decision through the neural network. disc rotation angle, but its interpretability is poor

Images