Adaptive cruise decision system and method, computer equipment and computer readable storage medium

Abstract

The invention discloses an adaptive cruise decision system and method, computer equipment and a computer readable storage medium, and the system comprises a decision model building module which is used for building a network decision model which comprises an upper-layer decision network, a bottom-layer execution network and a reward function; the initialization module is used for initializing environment parameters and acquiring an initial external state s0meta and an initial internal state s0controller of the environment; the decision execution module is used for selecting an initial sub-target parameter goali and initializing a reward parameter rimeta according to the external state parameter Simeta; sub-target completion conditions are judged, wherein the distance between the vehicle and the front vehicle is smaller than a preset minimum value or larger than a preset maximum value, and the speed of the vehicle is 0; if not, according to the internal state parameter sjcontroller, selecting an initial action aj, interacting with the environment to obtain an internal state parameter sj + 1controller, interacting with the environment to output a reward parameter rjcontroller, and storing a memory {sjcontroller, aj, rj, sj + 1controller, goali}; the rimeter is equal to the sum of the rimeter and the rj and the sum of the rj and the 1 controller; i is circulated from 1 to N until the sub-target is completed; and a memory {simeta, goali, rimeta, si + 1meta} is stored.

Description

technical field [0001] The present invention relates to the technical field of cruise decision-making, in particular to an adaptive cruise decision-making system, method, computer equipment and computer-readable storage medium. Background technique [0002] In recent years, the vigorous development of high-tech technology has driven the transformation of automobiles from pure means of transportation to a new generation of intelligent mobility spaces. Among the behaviors of intelligent driving, the adaptive cruise behavior in the lane is the most basic. It is responsible for giving appropriate speed control instructions from the starting point to the destination, which is executed by the control layer to guide the driving of the vehicle. [0003] During the driving of a smart car, in order to provide passengers with a comfortable driving experience and enable surrounding drivers to better understand the behavior of the car, the driving style of the adaptive cruise algorithm s...

AI Technical Summary

Problems solved by technology

The control strategy of the rule-based adaptive cruise control system is always in a fixed framework, and the strategy cannot be changed by different driver preferences and driving styles of different drivers, which reduces the adaptability of the rule-based adaptive cruise control algorithm and driver acceptance

[0005] In the design of adaptive cruise control algorithms based on learning, most researchers use single-layer deep reinforcement learning algorithms to build adaptive cruise algorithms, and use expert estimation methods to design reward functions. Execution” behavior mode, and the reward function designed by the expert estimation method does not anthropomorphize the reward function based on real driving data, so it is difficult to design an anthropomorphic adaptive cruise control algorithm based on this method

Images