Internet of Vehicles Dataset - Collision Detection and Safety

Dataset Overview

The Internet of Vehicles (IoV) Dataset focuses on vehicle safety and collision prevention through connected vehicle technologies. It captures comprehensive data from vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communications combined with onboard sensor measurements under various traffic scenarios.

Data Sources and Features

V2V Communication Data

• Vehicle Position: GPS coordinates with high precision (latitude, longitude, altitude)
• Velocity Vectors: Speed and direction of travel
• Acceleration/Deceleration: Braking and acceleration events
• Heading: Vehicle orientation and trajectory
• Neighboring Vehicles: Proximity data from nearby connected vehicles
• Communication Latency: V2V message transmission delays critical for collision avoidance

Onboard Sensor Data

• LiDAR/Radar: Object detection and distance measurements
• Camera Systems: Lane detection, traffic sign recognition, obstacle identification
• Ultrasonic Sensors: Close-range parking and low-speed maneuvering
• Wheel Speed Sensors: Individual wheel rotation rates
• Steering Angle: Driver input for trajectory prediction

Vehicle Status

• Brake Application: Brake pedal position and pressure
• Throttle Position: Acceleration input
• Gear Selection: Transmission state
• Turn Signal Status: Intention indicators
• Traction Control: ESC and ABS activation events

Traffic and Environmental Context

• Traffic Density: Number of vehicles in vicinity
• Road Conditions: Surface type, weather impact (dry, wet, icy)
• Time of Day: Lighting conditions affecting visibility
• Road Type: Highway, urban, rural classifications
• Speed Limits: Regulatory constraints

Collision Indicators - Target Variables

• Collision Risk Level: Low, moderate, high, imminent (multi-class classification)
• Time to Collision (TTC): Seconds until potential impact (regression target)
• Collision Occurred: Binary label for actual collision events
• Collision Type: Rear-end, side-impact, head-on classifications

Research Applications

Collision Avoidance Systems

• Develop ML models predicting collision risk seconds before occurrence
• Train systems providing automatic emergency braking or steering interventions
• Create early warning systems alerting drivers through visual/auditory signals

Autonomous Driving

• Decision-making algorithms for self-driving vehicles in complex traffic scenarios
• Path planning that proactively avoids high-risk situations
• Sensor fusion combining V2V communications with onboard sensors

Traffic Safety Analysis

• Identify road segments or conditions with elevated collision risks
• Evaluate effectiveness of safety interventions and road design changes
• Study human driving behaviors leading to near-miss events

V2V Protocol Optimization

• Assess impact of communication latency on collision prevention effectiveness
• Optimize message prioritization and transmission strategies
• Develop resilient systems handling intermittent connectivity

Machine Learning Tasks

• Classification: Predict collision risk categories or collision types
• Regression: Estimate time-to-collision for graduated response systems
• Sequence Modeling: Analyze temporal patterns leading to collisions using LSTM/GRU
• Reinforcement Learning: Train autonomous agents to navigate safely

IoV Ecosystem Research

Beyond individual vehicle safety, this dataset supports research into cooperative driving systems where vehicles coordinate behaviors through V2V communication to prevent collisions collectively, optimize traffic flow, and enable platooning.