MULTI-AGENT-BASED APPROACH FOR REAL-TIME COLLISION AVOIDANCE AND PATH REPLANNING ON CONSTRUCTION SITES

Collisions on construction sites are one of the major causes of fatal accidents. The complexity of equipment operations require detailed planning and better real-time control of the work. Research involving artificial intelligence in construction industry has been done to enhance communication between team workers and resolve distributed problems, for example, agent systems have been used for construction claims and dynamic rescheduling negotiation. However, little research has focused on real-time control of construction equipment operations using agents to improve safety on site. The present paper proposes a multi-agent-based approach to provide real-time support to the construction staffs. Collision avoidance is achieved by informing workers and equipment operators about potential collisions, and by providing replanning for equipment. A prototype system has been developed to and the functionalities of different agents are successfully tested.