The invention provides an unmanned aerial vehicle remote control algorithm based on sectional reinforcement learning. The unmanned aerial vehicle remote control algorithm comprises the following steps of definition of a state space and an action space, reward setting, a loss function, network structure design, a network training method, a network updating method and selection of a simulation environment. The unmanned aerial vehicle remote control algorithm based on sectional reinforcement learning provided by the invention has the following beneficial effects: a sectional reinforcement learning method inspired by curriculum learning is adopted to train a network, and the sectional reinforcement learning method divides a complex network into two sections for training, one section is speed control, and the other section is speed control; one is obstacle avoidance control, the training difficulty of the model can be reduced, the decoupled neural network is also beneficial to enhancing the interpretability of the neural network, the two networks can be operated separately when necessary, the adaptability is high, in addition, the PPO algorithm is adopted for network updating, and the training efficiency is further improved.
本发明提供一种基于分段式强化学习的无人机远程控制算法,包括以下步骤:状态空间和动作空间的定义、奖励设置和损失函数、网络结构设计、网络训练方法、网络更新方法、仿真环境的选择。本发明提供的一种基于分段式强化学习的无人机远程控制算法具有如下有益效果:采用一种由课程学习所启发的分段式强化学习的方法对网络进行训练,这种分段式学习的方法将一个复杂的网络分成两段来训练,一个是速度控制,一个是避障控制,可以减轻模型的训练难度,这种解耦的神经网络同时也有利于增强神经网络的可解释性,并且在必要的时候可以将两个网络分开运行,具有比较强的适应性,此外,本发明还采用了PPO算法进行网络更新,进一步提高了训练效率。
Unmanned aerial vehicle remote control algorithm based on sectional reinforcement learning
一种基于分段式强化学习的无人机远程控制算法
2024-01-16
Patent
Electronic Resource
Chinese
REINFORCEMENT LEARNING-BASED REMOTE CONTROL DEVICE AND METHOD FOR AN UNMANNED AERIAL VEHICLE
European Patent Office | 2020
|Reinforcement learning-based remote control device and method for an unmanned aerial vehicle
European Patent Office | 2023
|European Patent Office | 2021
|Positioning device, remote control terminal, unmanned aerial vehicle and unmanned aerial vehicle
European Patent Office | 2022
|