Training method and system of robot navigation obstacle avoidance model and related equipment

The invention provides a training method and system for a robot navigation obstacle avoidance model, and related equipment, relates to the field of mobile robots, and can improve the training effect of the robot navigation obstacle avoidance model and reduce the training time. The method comprises the following steps: in a training simulation environment, carrying out obstacle avoidance navigation training on a target pulse neural network model of the robot to obtain multiple groups of empirical data; and generating corresponding network state parameters according to the multiple groups of empirical data and a TD3 algorithm so as to update synaptic weights connected with neurons in the target spiking neural network model for multiple times to obtain the trained navigation obstacle avoidance model. 本申请提供一种机器人导航避障模型的训练方法、系统及相关设备，涉及移动机器人领域，可以提高机器人导航避障模型的训练效果，减少训练时长。其中，上述方法包括：在训练仿真环境下，对所述机器人的目标脉冲神经网络模型进行避障导航训练，以获取多组经验数据；根据所述多组经验数据和TD3算法生成对应的网络状态参数，以对所述目标脉冲神经网络模型中的神经元连接的突触权重进行多次更新，以得到训练后的所述导航避障模型。