AutoRepo: A general framework for multimodal LLM-based automated construction reporting

Ensuring the safety, quality, and timely completion of construction projects is critical, and construction inspections play an important role in achieving these objectives. Nevertheless, the predominantly manual approach to inspections frequently results in inefficiencies and inadequate information management. These methods often fail to provide a comprehensive and detailed assessment, resulting in regulatory oversight and potential safety hazards. To address this issue, this paper presents a novel framework named AutoRepo for automated generation of construction inspection reports. In AutoRepo, the unmanned vehicles efficiently perform construction inspections and collect scene information, while the multimodal large language model (LLM) is used to automatically generate the inspection reports. The framework was applied and tested on a real-world construction site, demonstrating its potential to speed up the inspection process, significantly reduce waste of resource, and generate high-quality inspection reports that meet regulatory standards. Thus, this research highlights the great potential of multimodal LLMs in revolutionizing construction inspection practices, signaling a significant leap forward towards a more efficient and safer construction management paradigm.