Evaluating a Multi-Modal Robotic Approach for Deconstruction Inspection: A POP-Based Comparison with a Manual Approach

Heyaojing Huang, Vaishnavi Thumuganti, Martin Fischer

Type: An evaluation-centric paper (focused on advancing methods and ideas for robot evaluation)

Keywords: Deconstruction inspection; Construction robotics

Abstract: Material reuse and waste reduction are gaining attention for their role in lowering carbon emissions and enabling net-zero buildings. Prior work has shown the potential of non-invasive deconstruction inspection using multi-modal sensing and machine learning to generate reuse recommendations. However, manual data collection remains operator-dependent, limiting consistency and repeatability. To address this, we integrated a quadruped robot for automated sensing and evaluated its feasibility using a product–organization–process (POP) framework. While the robotic workflow streamlines the organizational structure and simplifies the product setup, it still requires manual fine-tuning of viewpoints, as the robot lacks awareness of inspection targets and the ability to determine appropriate scanning behavior, and offers no real-time data quality feedback—relying instead on asynchronous review, which may lead to rework. Future work will focus on developing 3D object-level semantic perception, real-time quality assessment metrics, and mixed reality visualization to enhance human–robot collaboration, along with hardware adaptations and sensor integration to expand the robot’s multi-modal capabilities.

Submission Number: 10