The invention discloses an unmanned driving method based on a humanoid driving rule and an improved depth deterministic strategy gradient, and the method comprises the steps: carrying out the humanoiddriving behavior modeling in a mode based on a rule instead of a sample, and carrying out the unmanned driving strategy exploration based on a depth deterministic strategy gradient algorithm with theaddition of a humanoid behavior constraint. The improved unmanned driving strategy can output continuous human-like driving behaviors, has a better control effect, and can solve the problem that traditional imitation learning depends on samples.
本发明公开了一种基于类人驾驶规则与改进深度确定性策略梯度的无人驾驶方法,包括:使用基于规则而非基于样本的方式进行类人驾驶行为建模,并基于添加类人行为约束的深度确定性策略梯度算法进行无人驾驶策略探索。本发明所改进的无人驾驶策略能够输出连续的类人驾驶行为,具备更优秀的控制效果,并能解决传统模仿学习依赖样本的问题。
Unmanned driving method based on humanoid driving rule and improved depth deterministic strategy gradient
一种基于类人驾驶规则与改进深度确定性策略梯度的无人驾驶方法
2021-01-08
Patent
Electronic Resource
Chinese
HEV energy management method based on distributed depth deterministic strategy gradient
European Patent Office | 2021
|Electric vehicle ordered charging method based on depth deterministic strategy gradient algorithm
European Patent Office | 2023
|UNMANNED DRIVING SYSTEM, CONTROL DEVICE, AND UNMANNED DRIVING METHOD
European Patent Office | 2024
|UNMANNED DRIVING SYSTEM, CONTROL DEVICE, AND UNMANNED DRIVING METHOD
European Patent Office | 2024
|European Patent Office | 2024
|