Design And Performance Optimization Of Educational Robot's AI Algorithm Education System

Robots have developed from the initial industrial robots,military robots,service robots,to the field of entertainment robots and educational robots.Educational robots are currently mostly used for competitions,but less for teaching.At present,the development of artificial intelligence is in the ascendant,and the educational demand for artificial intelligence for young people is more and more urgent.Therefore,it is necessary to develop an educational robot with teaching functions in response to this demand.Therefore,this article designs an educational robot that can teach artificial intelligence algorithms.This robot is designed to help beginners,and through human-computer interaction,students can understand the ideas of artificial intelligence algorithms.The education system uses KNN algorithm for education,which is the simplest and effective classification algorithm,simple and easy to implement.Because of its simple and efficient characteristics,it is very suitable for teaching students who are new to artificial intelligence.Beginners can also understand and learn more easily during the teaching process.The robot can explain the KNN(k-NearestNeighbor)classification algorithm in detail.Through human-computer interaction,students can participate in each step of the algorithm.An example of KNN image classification is also provided,and students can understand the algorithm more thoroughly through the examples.Due to the stable performance of ESP32,which can realize wireless communication and has the characteristics of ultra-low power consumption,the KNN algorithm development and implementation of educational robots are implemented on the ESP32 embedded system.The main research contents of this article are as follows:(1)Design and implementation of KNN learning framework.KNN's learning framework is a combination of PC and ESP32.The KNN algorithm is a supervised learning algorithm,that is,using a set of samples of a known category to adjust the parameters of K,and finally a higher accuracy rate can be obtained.Because the real environment that educational robots face is relatively complicated,generally the pictures that need to be classified are presented in front of the robot vision system,and the information of the pictures to be segmented is extracted by the vision system.In this paper,two-pass algorithm is used for image segmentation,and the target is extracted from the complex background environment.There are three interfaces set on the PC,category selection interface,picture segmentation interface and classification interface,which are used to provide category information for training samples,collect training sets and display classification results.The picture segmentation interface will show the effect of segmentation and the effect of the obtained training samples.When designing the system,putting the training set acquisition process on the PC side can avoid a lot of calculations on the ESP32.After the training set is collected and divided,it is sent to the Flash corresponding to ESP32,and the classification calculation is performed on ESP32,so that the robot can be independently classified.The process of classification will allow students to participate in the calculation of distance,increase human-computer interaction,and achieve better teaching results.(2)Additional functions for system upgrades and hardware detection.Due to the rapid development of hardware,we have designed automatic detection and wireless upgrade functions of the system hardware.Hardware detection uses a multi-threaded method to detect the PID and VID of the ESP32.The upgrade uses the OTA method to upgrade the ESP32,download the bin file from the browser,and update automatically.(3)System performance optimization.Aiming at the problem that the segmentation speed is slow and the video freezes,this article analyzes how to improve the segmentation speed from the perspective of edge computing.The local computing task is sent to the edge server,and the high computing performance of the edge server is used to reduce the computing time.Because a large number of calculations are placed on the edge server,the local is only responsible for sending data,thus greatly reducing local energy consumption.This article discusses in detail how to allocate the resources of the edge server,compares the three different schemes of allocating the computing resources of the edge server according to the amount of data and priority,and the local computing.Finally,the simulation results are given to illustrate the edge Calculate the performance improvement of the band.