Event Log Dataset from a Decentralised Human-in-the-Loop Collision Avoidance System for Industrial Forklifts Based on Modified ORCA Algorithm

This dataset comprises a time-series event log exported from a MongoDB collection generated within the AA1 use case of the P2CODE project. The data originates from a human-in-the-loop collision avoidance system deployed on human-operated industrial forklifts in a logistics environment. The system architecture follows swarm intelligence principles, with each forklift equipped with a Raspberry Pi running a locally modified version of the Optimal Reciprocal Collision Avoidance (ORCA) algorithm. Rather than controlling vehicle motion autonomously, the system provides real-time advisory indications to human operators through a virtual traffic-light interface, where red signals indicate a stop, orange signals indicate caution, and green signals indicate normal operating conditions. The decentralised nature of the system means that no central coordination unit governs traffic flow; instead, forklifts periodically broadcast their position and motion context to nearby vehicles, enabling emergent coordinated behaviour through local interactions. The dataset records state transitions in the traffic-light indication system, capturing each change in collision risk assessment as a discrete event. Each record includes a timestamp in milliseconds, a unique forklift identifier, the traffic-light state at the moment of transition, and the vehicle's two-dimensional spatial coordinates. This dataset is intended for researchers investigating decentralised safety assistance systems, swarm-based coordination mechanisms in industrial settings, and the dynamics of human-in-the-loop decision support in collision avoidance scenarios.