In a complicated fusion device, the robot is at risk of colliding with its surroundings when it moves. In this article, a hybrid collision detection perceptron is studied to ensure the safety for both the fusion device and the robot. Primary component analysis is adopted as a preprocessor for extracting the features of the data. The stochastic gradient descent, Adam and Adagrad are used as classifiers with different learning rate methods. From the simulation results, it proves that this hybrid perceptron is more valid than the traditional detection method based on the dynamic model, and it reduces the complexity of robot modeling.

中文翻译：

---

机器人混合碰撞检测感知器在融合应用中的应用

在复杂的融合装置中，机器人在移动时存在与周围环境发生碰撞的风险。本文研究了一种混合碰撞检测感知器，以确保融合装置和机器人的安全。采用主成分分析作为提取数据特征的预处理器。使用随机梯度下降、Adam 和 Adagrad 作为具有不同学习率方法的分类器。仿真结果证明该混合感知器比传统的基于动态模型的检测方法更加有效，并且降低了机器人建模的复杂度。