ENHANCING VEHICLE SAFETY: AI-POWERED RISK LOGGING AND SEVERITY ASSESSMENT

Sammanfattning

Ensuring operational safety in heavy machinery, particularly construction equipment, is critical to minimizing risks and enhancing efficiency. This thesis investigates the potential of leveraging Artificial Intelligence (AI) for risk detection, classification, and logging within the construction equipment sector. A prototype system is developed utilizing sensor fusion techniques to integrate data from onboard sensors and AI-driven models, including Feed forward neural network (FFNN) and Long short-term memory (LSTM) models. The system identifies potential risks for an excavator, classifies them by severity, and logs this information systematically. Fuzzy logic is employed for data labelling, ensuring robust training of supervised AI models. The study explores critical parameters influencing risk classification ac curacy, such as sensor configurations and model architectures. Comprehensive testing reveals the system’s capacity to prioritize critical issues and provide actionable insights for safety enhancements. By focusing on Volvo Construction Equipment (Volvo CE) as a case study, this research delivers a scalable framework for deploying AI-based risk assessment systems across the construction equipment industry. The findings not only address existing technological gaps but also contribute to advancing safety and operational practices in increasingly automated and complex machinery ecosystems.