DESIGN OF MICROSLEEP DETECTION SYSTEM IN 32-BIT MICROCONTROLLER-BASED MOTORISTS WITH RANDOM FOREST METHOD

The number of motorcycle accidents has increased rapidly every year. Many occur due to drowsiness or fatigue because motorists force themselves to keep driving. The state of fatigue while driving is also known as microsleep. To overcome this problem, we propose a design of a prototype system that can be installed on the helmet of a motorized user so that the driver is more alert when driving a vehicle. This system utilizes machine learning technology with the Random Forest algorithm with two prediction results: prediction 1, which means the motorcyclist is tired, or prediction 0, which means the motorcyclist is in a normal state, embedded in the ESP32 microcontroller, and a tilt sensor that can detect signs of drowsiness in motorists. This system design will use the MPU6050 sensor to measure changes in the angle of the motorcyclist's head. The microcontroller will process the data obtained to identify head changes that indicate the possibility of drowsiness. If it occurs, the buzzer will beep as a warning to warn the driver to take a short break. The test results in drowsiness conditions with an angle of 10°–30° resulted in 100% accuracy, and normal conditions only at an angle of 0°–6° resulted in 100% accuracy. The result of the developed system is expected to reduce the number of accidents caused by drowsiness