ERI: Prediction-Enabled Safe and Productive Human-Robot Collaboration in Dynamic and Uncertain Construction Workspaces

该奖项全部或部分由《2021年美国救援计划法案》（公法117-2）资助。长期以来，建筑行业一直面临着低生产率和高伤亡率的挑战，而持续的劳动力老龄化和劳动力短缺进一步加剧了这一挑战。机器人建筑越来越被认为是一种很有前途的解决方案，可以将人类工人从危险和体力要求高的任务中解脱出来，从而提高生产率和职业健康与安全。然而，非结构化和动态的工作空间以及多样化和复杂的施工活动使得传统的预编程工业机器人在固定的工作环境中执行单一任务变得极其困难。该项目将通过开发新的方法来预测工人的行为意图，并考虑到工作环境和人类动态，自适应地规划机器人的运动，从而促进对动态环境中复杂建筑活动中人与机器人之间双向影响和相互作用的理解。开发的人机协作机制将为主动、自适应、安全和高效的机器人施工解决方案提供新的机会。从这项研究中获得的新知识将为复杂和动态环境中的人机协作提供见解，并可以推广到其他领域，如搜索和救援，农业和制造业。该项目还包括教育和外展活动，以促进学生的广泛参与，特别是那些来自代表性不足群体的学生，通过这些活动，年轻一代将被激励进入该行业，并获得与高科技合作的新能力，以提高行业和国家的竞争力。考虑到复杂的工人-机器人-工件-工作空间的相互作用，本研究将通过预测人的意图和运动，估计机器人的子任务，并生成最优运动轨迹，从而为动态和不确定建筑工作空间中安全高效的人机协作提供新的范例。本项目将以模板施工为工作情境，1)开发新的深度学习模型，整合人类行为数据和情境信息（如工作空间配置），预测工人的意图和运动以及相关的不确定性；2)开发新的分层控制机制，在多工序施工操作中推断机器人的任务，并考虑语义施工知识生成最优运动轨迹。预测人类动态和相关的不确定性，以实现协作安全和生产力。将进行广泛的实验，以收集与机器人合作时的多模式人类行为数据，并测试已开发技术的性能。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

This award is funded in whole or in part under the American Rescue Plan Act of 2021 (Public Law 117-2).Construction industry has long been challenged by low productivity and high rates of injuries and fatalities, while the persistent workforce aging and labor shortage further exacerbate the challenges. Robotic construction has been increasingly recognized as a promising solution to relieve human workers from dangerous and physically demanding tasks and thus improving productivity and occupational health and safety. However, the unstructured and dynamic workspaces and the diverse and complex construction activities make it extremely difficult to apply industrial robots that are traditionally pre-programmed to conduct a single task in a fixed working environment. This project will advance the understanding of the bi-directional influence and interaction between human and robot when collaborating in complex construction activities in dynamic environments by developing new methods to predict worker intention from their behavior and adaptively plan robot’s motion considering both job contexts and human dynamics. The developed human-robot collaboration mechanism will provide new opportunities for proactive, adaptive, safe, and productive robotic construction solutions. The new knowledge gained from this research will provide insights on human-robot collaboration in complex and dynamic environments, and can be generalized to other domains, such as search and rescue, agriculture, and manufacturing. This project also involves education and outreach activities to promote broad participation of students, especially those from underrepresented groups, through which younger generations will be motivated to enter the industry and trained with new capabilities to work with high technologies to increase competitiveness of the industry and the nation.This research will lead to a new paradigm for safe and productive human-robot collaboration in dynamic and uncertain construction workspaces by predicting human intention and motion, estimating robot subtasks, and generating optimal motion trajectory, considering the complex worker-robot-workpiece-workspace interaction. Using formwork construction as the working context, this project will 1) develop new deep learning models that integrate human behavior data and contextual information (e.g., workspace configuration) to predict worker’s intention and motion and associated uncertainties, and 2) develop new layered control mechanism to infer robot’s task in multi-process construction operation and generate optimal motion trajectory considering semantic construction knowledge, predicted human dynamics and associated uncertainties to achieve both collaboration safety and productivity. Extensive experiments will be conducted to collect multimodal human behavior data when collaborating with robots and to test the performance of developed technologies.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.