TY - JOUR
T1 - Multispecies hybrid bioinspired climbing robot for wall tile inspection
AU - Lin, Tzu Hsuan
AU - Chiang, Pin Chian
AU - Putranto, Alan
N1 - Publisher Copyright:
© 2023
PY - 2024/8
Y1 - 2024/8
N2 - This study introduces GLEWBOT (Gecko, Leech, and Woodpecker-inspired Robot), a cutting-edge, multispecies hybrid bioinspired robot engineered for inspecting exterior wall tiles. By integrating the adaptive locomotion techniques of leeches, the stability mechanisms found in gecko tails, and the percussive impact strategy of woodpeckers, GLEWBOT revolutionizes the inspection of wall surfaces. This innovative approach not only boosts inspection accuracy but also increases efficiency substantially. Constructed using 3D printing technologies, GLEWBOT combines lightweight design with structural integrity and integrates advanced microcontrollers for real-time, efficient inspection tasks. The robot features vacuum-based suction cup footpads for solid adhesion, minimizing detachment risks, and employs a compact machine learning model for instant on-site defect detection. This model, designed for edge computing, overcomes the hurdles of high computational demands. Additionally, GLEWBOT is equipped with an AI-powered acoustic recognition module, using a dedicated convolutional neural network (CNN) to assess tile integrity objectively. This significantly advances subjective manual inspections, offering a more reliable, objective analysis. Experimental trials involving systematic tapping across 26 tiles on a model wall validated GLEWBOT's defect detection capabilities. With an overall accuracy of 74.62%, precision of 75.4%, recall of 73.08%, and an F1-score of 74.22%, confirm GLEWBOT's potential as a reliable tool for identifying tile defects. Future work will focus on expanding GLEWBOT's versatility across different architectural materials and contexts while considering economic and regulatory factors for broader deployment. This research sets a new standard in robotic inspection for building maintenance and fosters interdisciplinary collaboration by illustrating the successful integration of bioinspired designs and artificial intelligence in structural inspections.
AB - This study introduces GLEWBOT (Gecko, Leech, and Woodpecker-inspired Robot), a cutting-edge, multispecies hybrid bioinspired robot engineered for inspecting exterior wall tiles. By integrating the adaptive locomotion techniques of leeches, the stability mechanisms found in gecko tails, and the percussive impact strategy of woodpeckers, GLEWBOT revolutionizes the inspection of wall surfaces. This innovative approach not only boosts inspection accuracy but also increases efficiency substantially. Constructed using 3D printing technologies, GLEWBOT combines lightweight design with structural integrity and integrates advanced microcontrollers for real-time, efficient inspection tasks. The robot features vacuum-based suction cup footpads for solid adhesion, minimizing detachment risks, and employs a compact machine learning model for instant on-site defect detection. This model, designed for edge computing, overcomes the hurdles of high computational demands. Additionally, GLEWBOT is equipped with an AI-powered acoustic recognition module, using a dedicated convolutional neural network (CNN) to assess tile integrity objectively. This significantly advances subjective manual inspections, offering a more reliable, objective analysis. Experimental trials involving systematic tapping across 26 tiles on a model wall validated GLEWBOT's defect detection capabilities. With an overall accuracy of 74.62%, precision of 75.4%, recall of 73.08%, and an F1-score of 74.22%, confirm GLEWBOT's potential as a reliable tool for identifying tile defects. Future work will focus on expanding GLEWBOT's versatility across different architectural materials and contexts while considering economic and regulatory factors for broader deployment. This research sets a new standard in robotic inspection for building maintenance and fosters interdisciplinary collaboration by illustrating the successful integration of bioinspired designs and artificial intelligence in structural inspections.
KW - Acoustic recognition
KW - Climbing robot
KW - Convolutional neural network (CNN)
KW - Defect detection
KW - Edge computing
KW - Exterior wall tiles inspection
KW - GLEWBOT
KW - Multispecies hybrid bioinspired robotics
KW - Tiny machine learning
UR - http://www.scopus.com/inward/record.url?scp=85192071776&partnerID=8YFLogxK
U2 - 10.1016/j.autcon.2024.105446
DO - 10.1016/j.autcon.2024.105446
M3 - 期刊論文
AN - SCOPUS:85192071776
SN - 0926-5805
VL - 164
JO - Automation in Construction
JF - Automation in Construction
M1 - 105446
ER -