The invention discloses an automatic driving behavior decision-making method for view shielding. The method specifically comprises the following steps: step 1, acquiring scene information; 2, obtaining time, position information and scene element information, and deducing to obtain potential risks of a view shielding scene; 3, fusing the data obtained in the step 1 and the data obtained in the step 2; 4, generating a decision scheme based on rule mapping; 5, generating a decision scheme and a driving behavior based on the deep reinforcement learning model; and step 6, selecting an optimal driving decision. According to the method, the non-direct observation information in the view shielding scene is mined, the mode that the non-direct observation information and the direct observation information are subjected to numerical coding is used as input of the decision method, and compared with a deep reinforcement learning decision model which independently takes a direct observation value as input, the method has better convergence and decision accuracy.
本发明公开了用于视野遮挡的自动驾驶行为决策方法,具体按照以下步骤:步骤1、获取场景信息;步骤2、获取时间、位置信息及场景元素信息推断得到视野遮挡场景的潜在风险;步骤3、将所述步骤1与步骤2获取的数据融合;步骤4、产生基于规则映射的决策方案;步骤5、产生基于深度强化学习模型的决策方案与驾驶行为;步骤6、最优驾驶决策的选择。本发明通过挖掘视野遮挡场景中的非直接观测信息,将其与直接观测信息进行数值编码的方式作为决策方法的输入,相较于单独以直接观测值作为输入的深度强化学习决策模型具有更好的收敛性和决策的准确性。
Automatic driving behavior decision-making method for view shielding
用于视野遮挡的自动驾驶行为决策方法
2023-12-26
Patent
Electronic Resource
Chinese
IPC: | 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 |
Automatic driving vehicle behavior decision-making system and method
European Patent Office | 2023
|Automatic driving behavior decision-making system and method based on driving preference
European Patent Office | 2022
|European Patent Office | 2023
|Automatic driving behavior decision-making method based on deep reinforcement learning
European Patent Office | 2020
|