The invention relates to an automatic driving human-like safety self-evolution method and system based on data mechanism fusion, and the method comprises the steps: extracting real human driving data features from historical experience data, and carrying out the iterative extraction of a target function corresponding to the decision-making habit and planning habit of a driver through a maximum entropy inverse reinforcement learning algorithm; sampling from a traffic environment in real time, obtaining environment information, constructing an experience return visit pool including a current state, an action, an award and a next moment state, constructing a Q value neural network, extracting data from the experience return visit pool, performing iterative updating on the Q value neural network, and obtaining anthropomorphic constraints; and establishing a vehicle model, substituting the environment information at the current moment, constructing a vehicle actuator constraint, and performing search solution by combining the vehicle model, the personification target function and the personification constraint to obtain vehicle control information. Compared with the prior art, the self-driving automobile has self-learning performance and adaptability, and safe, efficient and comfortable driving can be achieved.
本发明涉及一种基于数据机理融合的自动驾驶类人安全自进化方法和系统,方法包括:从历史经验数据中,提取真实人类驾驶数据特征,通过最大熵逆强化学习算法,迭代提取出与该驾驶员决策习惯和规划习惯的目标函数;实时从交通环境中采样,获取环境信息,构建包括当前状态、动作、奖励和下一时刻状态的经验回访池,构建Q价值神经网络,从经验回访池抽取数据,对Q价值神经网络进行迭代更新,获取拟人化约束;建立车辆模型,并代入当前时刻的环境信息,构建车辆执行器约束,结合车辆模型、拟人化目标函数和拟人化约束进行搜索求解,获取车辆控制信息。与现有技术相比,本发明使得自动驾驶汽车具有自学习性和适应性,能实现安全、高效、舒适地驾驶。
Automatic driving human-like safety self-evolution method and system based on data mechanism fusion
基于数据机理融合的自动驾驶类人安全自进化方法和系统
2023-06-23
Patent
Electronic Resource
Chinese
IPC: | G05D SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES , Systeme zum Steuern oder Regeln nichtelektrischer veränderlicher Größen / B60W CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION , Gemeinsame Steuerung oder Regelung von Fahrzeug-Unteraggregaten verschiedenen Typs oder verschiedener Funktion / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen |
European Patent Office | 2023
|Automatic driving safety analysis system and method based on big data
European Patent Office | 2023
|Safety architecture for automatic driving, automatic driving system and vehicle
European Patent Office | 2024
|Vehicle automatic driving control method and system based on fusion
European Patent Office | 2022
|