Abstract
Machine learning for autonomous can be done by recording the displacement and condition of the vehicle through manual control by humans and modeling the data.The research proposes designing a data collection system for tree-based data modeling on Internet of Things (IoT) based autonomous electrical vehicles (EV).The system consists of four ESP32 cameras with servos mounted on the left, right side of the car mirror, front (dashcam), and rear.The system is also equipped with an Arduino Nano connected to GPS, a gyroscope, and four proximity sensors.Arduino nano is connected via serial software to the Wemos D1 mini, which is connected to a relay module to control lights and wipers and is equipped with an LDR sensor.Data collected via the internet (wifi) will be formed in treebased data modeling for future genetic programming machine learning algorithms.System evaluation includes Quality of Service (QoS) data communication, statistical data collected, and electrical IoT devices built.Based on testing using an intelligent car chassis in an environment still affordable by wifi, it produces an average delay of 0.02 s and a PDR of 99.87%.The highest correlation matrix archived as 0.872 for longitude, latitude, and gyro data in detecting vehicle turns.The electricity evaluation result consists of average power consumption of 0.344 W for the ESP32 camera, 0.663 W for the Arduino nano, and 0.291 W for the Wemos d1 mini.In the future, testing will be carried out using an actual EV on a real track and in data communication outside of wifi.