Multi-Modal Autonomous Driving Sensor Dataset (MM-ADS)

A multi-modal autonomous vehicle sensor dataset was collected to support the development and evaluation of the proposed AI-based autonomous driving framework. The dataset consists of 50,000 time-synchronized samples, divided into 70% training data (35,000 samples) and 30% testing data (15,000 samples). Each sample corresponds to a single time step recorded during vehicle operation under controlled driving scenarios.The dataset includes sensor-only measurements, excluding RGB image data, and integrates information from LiDAR, radar, ultrasonic sensors, GPS, and IMU, along with ego-vehicle dynamics such as speed and yaw rate. LiDAR and radar readings are provided in structured numerical form to capture spatial and motion characteristics efficiently, while GPS and IMU data support localization and vehicle state estimation. Ultrasonic sensor data enable short-range obstacle detection.Data were collected across urban and highway environments under varying weather conditions (clear, rain, fog) and lighting settings (day and night) to ensure diversity and robustness. All data are stored in CSV format, with each row representing a synchronized sensor observation.This dataset is used to train and evaluate sensor fusion, decision-making, energy-aware route optimization, and explainable AI models within the proposed autonomous vehicle framework, ensuring reproducible and efficient experimentation.