There are a large number of worker-machine collaboration scenarios in civil engineering construction sites,and human-machine collision is a major safety hazard in these scenarios.Traditional safety management methods focus on real-time hazard detection and post-accident investigations,lacking the capability to issue early warnings for human-machine collisions.A preemptive human-machine collision warning method based on the prediction of workers'trajectories is proposed.By analyzing working statuses and movement states of workers,their future movement trajectories are predicted.A human-machine collision risk function is introduced to calculate risk values based on the predicted trajectories,these risk values are then used to trigger early warnings.The method was validated in a residential building project in Shanghai.The results demonstrated that appling method could predict workers'future 2s trajectories with an average displacement error of 83cm and a final displacement error of 24cm.It was also able to issue warnings 2.1s before an impending human-machine collision accident.
维普
万方数据