语音示教+自主巡航智能垃圾分类机器人的研究与设计

制造一种能够语音示教,并能够自主巡航的智能垃圾分类机器人。该机器人可以通过CNN来识别各种垃圾,结合其他传感器提高识别准确率,并分类投放到相应的垃圾桶内,还可以通过屏幕和语音反馈给用户。该机器人可以通过双目视觉和超声波测距模块,绘制环境地图,进行避障和规划巡航路线。另外可以加入NLP,可实现语音控制。使用Google开源的神经网络框架Tensorflow搭建卷积神经网络进行图像识别。采用ReLU作为激活函数,能在反向传播时避免梯度消失。从测试结果来看,目前大部分垃圾的识别准确率在80%左右。

An intelligent garbage sorting robot which can instruct by speech and cruise autonomously is manufactured.The robot can identify all kinds of garbage through CNN(convolutional neural network),improve the recognition accuracy by combining with other sensors,classify and put into the corresponding garbage can,and also give feedback to users through screen and voice,telling them what kind of garbage they put in.The robot can use binocular vision and ultrasonic ranging modules to map the environment,avoid obstacles and plan cruise routes.In addition,NLP(natural language processing)can be added to achieve voice control.Google's open source neural network framework Tensorflow is used to build a convolutional neural network for image recognition.The ReLU(Rectified Linear Unit)can be used as the activation function so as accelerate the learning speed of neural network and avoid the disappearance of gradient in back propagation.According to the test results,most of the garbage currently has an accuracy rate of around 80%.