Abstract

The advent of Industry 4.0 heralds a transformative era for the Architecture, Engineering, and Construction (AEC) industry, characterized by the digitization of manufacturing processes leveraging industrial robot platforms. Unlike other sectors where industrial robots handle repetitive tasks that need to be simulated only once, the AEC industry involves customized object manufacturing, necessitating simulation and testing for each unique process. This bespoke nature underscores the necessity for efficient and reliable manufacturing simulation tools.

Robot simulation in the AEC industry typically relies on inverse kinematics calculations to derive robot axis values, determining the Tool Center Point (TCP). However, inverse kinematics analysis itself is highly complex, often requiring exploration of multiple solutions and constrained by specific structural conditions. Therefore, identifying suitable Machine Learning (ML) candidate solutions to address this challenge constitutes a primary objective of this paper. The proposed solution involves comparing the predictive accuracy of two ML techniques and validating them based on the nominal kinematic model of the robot.

In summary, with the arrival of Industry 4.0, the AEC industry is actively exploring new paradigms of digital manufacturing processes. Through the adoption of advanced robot simulation techniques and machine learning methods, the AEC sector aims to enhance production efficiency and achieve customized manufacturing capabilities.