This paper discusses how to improve the accuracy of navigation for home service robot based on the deep learning and machine learning. First, the crawling programing is applied to collect enough images of fridge and washing machine on the web; a deep learning framework is proposed that can distinguish fridge and washing machine more accurately. Following, the data come from the robot operating system topics are collected and cleaned, the linear regression, decision tree, and linear SVR algorithms are applied and compared to predict the power consumption of the robot, and a conclusion is obtained that liner movement will consume more power, which provides a reference for the path planning of the robot. Lastly, the conclusions are proposed that a novel methodology is applied to distinguish different home appliances, which is useful for the accurate navigation of the robot; the liner movement will consume more power compared to turning left or right, which supplies a reference for the optimized path planning for the robot.