SNU Innovations(vol.08)

• Designing Human-Robot Interfaces for In-Situ Improvisations
• Integrating XR for Semi-Autonomous Teleoperation of Construction Robots
• Developing RL-Based Navigation Methods for On-Site Monitoring

Prof. Changbum "Ryan" Ahn co-leads the Construction Engineering and Management Lab(SNUCEM) at the Department of Architecture and Architectural Engineering. Prof. Ahn is also leading the Smart Construction AI Research Center at the SNU AI Institute(AIIS), which aims for multidisciplinary research centered on artificial intelligence(AI) and robotics in construction.

Recently, the construction industry has been experiencing significant changes, particularly with the introduction of robots onto construction sites. Robots, previously limited to manufacturing shop floors, are now being utilized in off-site work such as modular construction in factories and in on-site tasks such as overhead drilling, wall painting, and autonomous excavation. According to ABB Robotics, 81 percent of construction businesses are likely to introduce or increase their use of robots by 2030¹⁾. This shift is driven not only by the pursuit of increased productivity and safety but also by challenges such as the aging of skilled workers and imbalances in labor supply and demand.

However, integrating robots into dynamic and unstructured construction environments presents significant challenges. One major issue is the discrepancies between the as-planned model and the as-built workpiece on site—a phenomenon widely known in construction—which can lead to errors that impact the quality of robot deployment. Additionally, construction tasks are often non-repetitive and have long planning horizons, which can make programming robots for these tasks particularly difficult. Lastly, and perhaps most importantly, there are varied perceptions of robotization among construction stakeholders. Some are hesitant to adopt robotics due to concerns about job loss or safety issues, leading to resistance to change and slow adoption of robotics in the industry ²⁾. In this context, developing an effective human-robot collaboration system is essential—a challenge that Prof. Ahn and his team at SNUCEM are addressing.

To realize this vision of viable human-robot collaboration, the team is developing human-centered interfaces and systems tailored to construction scenarios. One example is a multimodal interface that allows workers to make in-situ improvisations with overhead drywall cutting robots ^3,4). Using this interface, novice users with no expertise in robotics can operate the robots on-the-fly with 61% higher accuracy compared to conventional controllers used in the construction robots, without requiring any building information modeling(BIM) or simulation models of the environment in advance.

Another area of research involves enhancing teleoperation in construction using immersive visualization and control techniques. By adopting technologies such as virtual reality(VR) and augmented reality(AR), the team aims to improve safety and precision during the remote operation of construction robots or equipment, while carefully balancing cognitive load to maintain productivity. Recently, SNUCEM has been focusing on integrating extended reality(XR) for bimanual teleoperation and continual learning of multi-task, work-at-height construction robots^5,6).

In addition to focusing on robots that perform construction tasks, the team is also investigating methods to use robotic systems to monitor workers' safety on construction sites. One of the team's projects involves developing reinforcement learning-based navigation for mobile robots to navigate environments adaptively. This approach mitigates visual occlusion during monitoring, increasing the accuracy of detecting personal protective equipment(PPE) by 53% compared to existing safety monitoring systems that rely on fixed or mobile surveillance cameras.

The implications of SNUCEM's research are significant for the construction industry. While there are still several hurdles to overcome, the integration of robotics certainly holds potential to transform how we build our cities, infrastructure, and homes. By developing the aforementioned core technologies, SNUCEM is not only leading the research trends in construction robotics but also actively shaping how the industry approaches automation.

"Our aim is to empower construction workers to operate robots themselves, without the need for specialized operators," Prof. Ahn explains. "We're focused on developing robots that work alongside workers, not replace them. It's about fostering collaboration in which technology enhances human capabilities with robotic capabilities."