Abstract

The purpose of the autonomous racecar is to design and develop a micro-racecar with autonomous navigation in order to go through a simulated track by itself. To produce the autonomous racecar, it will  re-engineer a standard RC car by implementing an NVIDIA Jetson Xavier NX development kit and designing a power distribution system for all of the components. When powering the system, 2 separate power supplies will be incorporated for the powertrain motor with various components and computing board respectively. The Jetson will use data collected in the form of frames to train a Convolutional Neural Network model. The car’s Autonomous Control mode will be created by utilizing the Neural Network to predict values for the car’s throttle and wheel direction based on real-time video comparison. The autonomous car control will be programmed using Python and the Neural Network implementation will be done by applying the TensorFlow library. The car will also have a manual control mode in which a user can control the car using an Android Application built using Android Studio in addition to setting it to autonomous mode.