AWS Deepracer is a fully autonomous 1/18th scale race car driven by reinforcement learning , produced by AWS (Amazon Web Services). Users can train the autonomous driving model through AWS and then compete. We hope that AWS DeepRacer can be modified to make it more interesting, not just for competitions. It is also a great robotic tool for understanding machine learning. In our project we modified the AWS Deepracer to be an “AE86”, we connect the vehicle to the cloud through AWS IoT service, use the gaming driving racing wheel connected to the computer for the controller, cooperate with the camera on the vehicle to transmit the real-time picture to the computer for real-time driving. Moreover, to help an AWS DeepRacer successfully complete a race, we use the training model powered by Amazon SageMaker. However, the vehicle appears to perform differently when running in the real world. Therefore, we are developing a tool that is able to analyse any AWS DeepRacer’s performance through an AI service (by Amazon Rekognition Custom Label).