Like many other professions which depend on the repetition of complex motor patterns, masonry and construction often lead to injury, fatigue, and movement-related damage to joints and ligaments.
As a potential solution, researchers at the University of Waterloo in Ontario, Canada have recently proposed to use artificial intelligence (AI) to instruct novice trainees in safer and more ergonomic working techniques.
“The people in skilled trades learn or acquire a kind of physical wisdom that they can’t even articulate,” said Carl Haas, Professor of Civil and Environmental Engineering at Waterloo. “They’re basically doing the work twice as fast with half the effort, and they’re doing it with higher quality.”
Given the inability of most veteran professionals to explain their approach, Haas and colleagues hooked up 21 volunteers at different levels of experience with Xsens MVN motion-tracking suits and had them build a brick wall.
The data confirmed that expert masons put the least amount of stress on their musculoskeletal system, while performing work at a faster rate than their less experienced colleagues.
Interestingly, expert masons were also found to avoid following the standard advice given to novice trainees, instead relying on seemingly intuitive movement patterns, such as swinging more than lifting, and refraining from bending their backs.
Next, the researchers used the sensors to capture the participants’ movements, and fed the resulting data into a Support Vector Machine (SVM) algorithm to find specific patterns of motion.
Approximately 70% of the data was used for training the algorithms, while the remaining 30% was employed for testing.
The resulting “pose books” could be used in devising systems to train novice workers. “The classification method and the poses identified contribute knowledge to help develop affordable mason training systems that utilize IMU [inertial measurement units] and video feedback to improve health and productivity of apprentice masons,” wrote the authors in their paper.
Such systems could be based around suits with in-built motion sensors to give trainees immediate feedback on their movements.
The paper was published in the journal Automation in Construction.