Driver Behavior Dataset

Dataset Overview

Dataset Name

Driver Behavior Dataset

Dataset Description

The dataset gathers smartphone sensor measurements (accelerometer, linear acceleration, magnetometer, gyroscope) captured during predefined driving events using an Android application.

Experimental Setup

• Vehicle: 2011 Honda Civic
• Smartphone: Motorola XT1058, Android 5.1
• Sensor sampling rate: 50–200 Hz
• Drivers: 2 experienced drivers (≥15 years)
• Weather & road: clear, dry asphalt

Data Collection Method

• Smartphone fixed to the windshield, untouched during the drive.
• Android app logs sensor data with millisecond timestamps.
• One driver operates the vehicle while a rear‑seat assistant records video and cues the execution of specific events.

Driving Event Types

Dataset Structure

Each trip folder contains six files:

• aceleracaoLinear_terra.csv
• acelerometro_terra.csv
• campoMagnetico_terra.csv
• giroscopio_terra.csv
• groundTruth.csv
• viagem.json

groundTruth.csv holds start/end timestamps and event labels.

Related Works

• Driver behavior profiling: An investigation with different smartphone sensors and machine learning
• Exploiting the Use of Recurrent Neural Networks for Driver Behavior Profiling
• Análise de Perfil de Motoristas: Detecção de Eventos por meio de Smartphones e Aprendizado de Máquina