Abstract: Against the background of intelligent transformation in the construction industry, the reform of building systems brings diversified optimization possibilities and creative potential. To respond to the opportunities and challenges of human-machine symbiosis in future, the adaptive control concept removes static and presetting factory patterns with consideration for the complexity of building design and dynamics of construction scenes, integrating building design and intelligent construction and thus enabling systematic control thinking and cognition oriented to architects. Leading research and practices on adaptive construction were summarized via literature review, which included design needs and construction orientations to adapt to unconventional design schemes and construction processes, adaptation strategies for restrictions and interferences in the construction environment, and the specific performance and strain behaviors of different building materials. As a result, the evolutionary development of intelligent design and construction under adaptive control in the last decade were determined by analyzing typical practice cases of adaptive robotic construction theory in the intelligent construction field. Based on a multi-scale case study, the feedback cycle mechanism of adaptive control in key application technologies was determined, such as precision control during manufacturing of building components, dynamic environmental response during field construction, and building material adaption and strain prediction based on machine learning. Firstly, the basic control idea in the adaptive construction process with dynamic perceptions, calculation, and corrections was analyzed through a case study of cold-formed steel rod processing from the University of Michigan. This study pointed out that adaptive construction has advantages in reducing processing error accumulation and further enables compensation for construction errors. Secondly, the application of adaptive control to material selection on construction sites and field intelligent construction models was interpreted through an in-situ construction project at the Swiss Federal Institute of Technology, Zurich. This project not only realizes automatic perception and allows self-orientation in the working environment by using the construction site as the coordinate reference system, but also implements real-time monitoring of the adaptive construction workflow. Thirdly, the potential for an offline forecasting strategy based on machine learning to replace the practical construction scenes of construction engineering and effectively reduce time cost for precision control and error compensation was elaborated. Finally, limitations and potential problems of intelligent construction under adaptive control were discussed from the perspectives of complexity, continuity, and safety. Ultimately, the role of high-intensity data flow in constraints against computation time, storage space, and transmission cost as well as safety risks of human-machine cooperation were identified. The global influences of different designers on technical selection and strategic setting of local links during adaptive design construction processes, including the core nodes of "design-response-construction-perception", were discussed. Further requirements for self-perception, self-construction, and self-combination of buildings as well as systematic control thinking and global consciousness of future architects were proposed.

Key words:  , adaptive control, building intelligence, robotic fabrication, concept cognition, applied practice 

摘要： 在建筑行业智能化转型的背景下,建造体系的革新为建筑实践提供了多样的优化可能性和创造潜力。针对建筑设计建造的复杂性和建造场景的动态性特点，通过对于当前应用建筑智能化领域的机器人建造自适应原理相关内容及其实践案例的研究解析，建立起面向建筑师的系统控制思维和认知。从建筑构件制造、建造环境响应、材料性能测试三个方面的细分研究入手，梳理了近十余年来以自适应控制导向下的建筑智能化设计建造的演化发展脉络；结合国内外多尺度案例分析，厘清建筑建造过程中在加工精度把控、动态环境交互和材料行为模拟等应用技术研究中的反馈循环机制；总结了包括“设计-响应-建造-感知”等核心节点在内的自适应设计建造流程以及潜在影响，并展望其未来的发展前景。

关键词: 自适应控制, 建筑智能, 机器人建造, 概念认知, 应用实践