Vehicle hardware-in-the-loop simulation training system and method based on deep reinforcement learning

Abstract

The invention belongs to the field of unmanned vehicle simulation testing, and particularly relates to a vehicle hardware-in-the-loop simulation training system and method based on deep reinforcement learning. The system comprises a virtual simulation unit and a real vehicle controller, wherein the virtual simulation unit comprises a simulation vehicle module, a simulation control module, a real vehicle track database, a simulation sensor module, a reset interface and a start-stop interface. The real vehicle controller and the virtual simulation scene are subjected to joint simulation training, effective operation of the deep reinforcement learning movementplanning model in a real vehicle unmanned driving system is achieved, and therefore the problem that the deep reinforcement learning model cannot be effectively transplanted from a virtual simulation vehicle to an unmanned real vehicle is solved.

Description

technical field [0001] The invention belongs to the field of unmanned vehicle simulation testing, in particular to a vehicle hardware-in-the-loop simulation training system and method based on deep reinforcement learning. Background technique [0002] The motion planning module is an important technical link for the vehicle to realize unmanned driving. It plays an important role in the unmanned driving perception-motion (planning)-control-execution technology chain, which is equivalent to the driver's brain. In the past, research on motion planning for unmanned vehicles mainly used rule-based methods, model-based methods, and deep learning data "feeding" methods. Since the rule-based and model-based methods can only perform parameter modeling for specific simple scenes, they are not suitable for complex and changeable urban driving scenes. The motion planning model based on deep learning not only needs to collect a large amount of scene data for learning, but also can guaran...

AI Technical Summary

Problems solved by technology

[0004] In view of the above problems, the present invention proposes a hardware-in-the-loop simulation training system and method for vehicle motion planning based on deep reinforcement learning. The effective operation of the unmanned driving system, so as to solve the problem that the deep reinforcement learning model cannot be effectively transplanted from the virtual simulation vehicle to the unmanned real vehicle

Images