Full Text
TYPE SOO MINE: A BROWSER-BASED SMART GLASSES PLATFORM FOR REMOTE SUPPORT AND KNOWLEDGE SHARING IN MUNICIPAL WATER INFRASTRUCTURE
ABSTRACT
Japan's water and sewerage infrastructure faces a compounding crisis: declining technical staff, retiring veteran engineers, aging facilities, and increasingly complex emergency response. In small municipalities, field operations often depend on the experience and judgment of a limited number of technical personnel. This study investigates how a browser-based smart glasses platform can transform real-time remote assistance into reusable organizational knowledge for municipal water infrastructure operations. We present Type Soo Mine (TSM), a technology-sharing smart glasses platform jointly developed by Soo City, Kagoshima Prefecture, and a private XR development company, positioned as Japan's first browser-based infrastructure DX model integrating field sharing, recording, and education. TSM adopts a dedicated-application-independent, browser-based architecture: field workers wearing smart glasses stream first-person video and audio to remote supervisors, equipment manufacturers, and maintenance contractors, who provide simultaneous multi-party guidance through speech-to-text communication, annotated screenshots, and document display on the glasses. The system is designed to automatically record video, audio, chat, and operation logs from every session; these records will be processed using AI for key-scene extraction, transcription, and work-process summarization, then restructured into reusable training content with planned integration into the Moodle learning management system. Rather than treating remote support as a temporary communication function, TSM captures the judgment process of experienced personnel, what they observe, prioritize, and decide, and converts it into organizational knowledge. A pre-implementation survey of 146 respondents found 80.6% willing to use the system and 85.2% supporting the initiative. A pilot deployment with ten smart glasses units is underway: the time less-experienced staff can operate independently under remote support has tripled, and re-inspection rates have decreased by 50%. TSM is designed as a replicable model for small municipalities nationwide, with potential applications in construction, agriculture, and disaster response.
INTRODUCTION
Municipal water and sewerage services in Japan are maintained by a shrinking technical workforce: veteran engineers who can judge the condition of pumps, valves, and pipelines from sound, vibration, and visual cues are retiring while the facilities they maintain grow older. When the few experienced individuals in a small municipality are unavailable, inspection quality and emergency response both suffer.
Remote support tools and training systems each address part of this problem, but separately: video calls connect a field worker to a remote expert, yet the exchange disappears once the call ends, while training systems depend on someone authoring content after the fact, which rarely happens in understaffed utilities.
This paper presents Type Soo Mine (TSM), a browser-based smart glasses platform jointly developed by Soo City, Kagoshima Prefecture, and a private XR development company, positioned as Japan's first browser-based infrastructure digital transformation (DX) model integrating field sharing, recording, and education in one workflow. Unlike systems designed primarily for skills training, TSM starts from remote work support in actual operations and treats every support session as a recording to be reused.
We address the following research question: how can a browser-based smart glasses platform transform the real-time remote support that small municipal utilities must perform daily into structured, reusable organizational knowledge and training content, without requiring separate documentation effort from already understaffed field teams?
RELATED WORK AND POSITIONING
Smart glasses have been applied to remote assistance in manufacturing, plant maintenance, and healthcare, typically through dedicated applications tied to specific devices; for small municipalities with limited IT staff and mixed device environments, such installation and account management are costly. A second gap concerns what happens after the call: prior deployments generally treat the session as transient communication, and few mechanisms capture veteran judgment during routine work before it is lost at retirement. TSM is positioned at this junction: remote support as the capture point, training content as the downstream product.
SYSTEM DESIGN
Architecture. TSM is dedicated-application-independent: field workers wear smart glasses that stream first-person video and audio through a browser-based platform, and remote participants join from personal computers, smartphones, or tablets without installing software, reflecting the reality of small municipalities and partners whose device environments cannot be standardized.
Multi-party remote support. A session connects the field worker simultaneously with remote supervisors, equipment manufacturers, and maintenance contractors; guidance flows through speech-to-text communication, annotated screenshots, and document display on the glasses, so a worker in a noisy pump station can receive instructions visually, hands-free.
Recording and knowledge conversion. The system is designed to automatically record video, audio, chat, and operation logs from every session, with the intent of capturing the judgment process of experienced personnel, what they observe, prioritize, and decide, rather than only the procedures they perform.
AI processing pipeline. The AI processing pipeline is currently in the design phase and follows three planned stages. First, automatic speech recognition converts session audio into searchable transcripts. Second, a scene-segmentation step identifies candidate key moments in the first-person video, using cues such as annotation events and dialogue activity rather than manual review of full recordings. Third, transcripts and identified scenes are summarized into structured work-process descriptions, from which draft captions and assessment questions are generated. Because these outputs inform field safety judgment, all AI-generated content is intended to be reviewed by a domain expert before publication to learners through the Moodle learning management system.
DEPLOYMENT
Before implementation, we surveyed 146 respondents connected to the city's water and sewerage operations; 80.6% expressed willingness to use the system and 85.2% supported the initiative, indicating that field acceptance was not a major obstacle.
A pilot deployment using ten smart glasses units is underway in water and sewerage inspection operations in Soo City; veteran staff and external specialists support less-experienced workers remotely, and first-person records are already being captured digitally and reused within the organization.
PRELIMINARY RESULTS
Two preliminary indicators are reported from the ongoing pilot. First, the time that less-experienced staff can operate independently under remote support has roughly tripled compared with prior practice. Second, re-inspection rates have decreased by about 50%: problems are more often resolved correctly on the first visit when remote specialists can see what the field worker sees.
We treat these figures with caution, as they come from a single municipality, a limited number of units, and operations still adjusting to the platform. Quantitative evaluation of latency, first-response accuracy, and training-material effectiveness is in progress and will be addressed in future work.
DISCUSSION
We draw three initial observations from the pilot. First, browser-based architecture mattered: external manufacturers and contractors joined sessions without installation steps, shortening the path from problem found to specialist consulted. Second, positioning recording as a by-product of support work avoided the authoring bottleneck that stalls many knowledge-management efforts; records accumulate because the work itself requires the session. Third, field staff accept being recorded when the immediate benefit, help while working, is concrete.
We also acknowledge several limitations: the evidence base is one municipality and ten units; the AI pipeline and Moodle integration are designed but not yet operating at scale; the two headline indicators lack controlled baselines; and environmental conditions such as humidity, lighting, and network coverage have not yet been analyzed systematically.
Further work will expand data collection across sites, personnel, and operational conditions to examine generalizability, and will explore transfer to adjacent fields such as construction, agriculture, and disaster response.
CONCLUSION
We presented TSM, a browser-based smart glasses platform linking remote work support, session recording, and training-content generation in a single workflow. A pilot with ten units in Soo City indicates that less-experienced staff can work independently roughly three times longer under remote support and that re-inspection rates have fallen by about half, while first-person records accumulate for reuse. These are early results and broader validation is in progress; even so, for small utilities facing staff decline, remote support and knowledge preservation need not be separate investments, since a browser-based platform can serve both from the same daily work.
ABSTRACT
Japan's water and sewerage infrastructure faces a compounding crisis: declining technical staff, retiring veteran engineers, aging facilities, and increasingly complex emergency response. In small municipalities, field operations often depend on the experience and judgment of a limited number of technical personnel. This study investigates how a browser-based smart glasses platform can transform real-time remote assistance into reusable organizational knowledge for municipal water infrastructure operations. We present Type Soo Mine (TSM), a technology-sharing smart glasses platform jointly developed by Soo City, Kagoshima Prefecture, and a private XR development company, positioned as Japan's first browser-based infrastructure DX model integrating field sharing, recording, and education. TSM adopts a dedicated-application-independent, browser-based architecture: field workers wearing smart glasses stream first-person video and audio to remote supervisors, equipment manufacturers, and maintenance contractors, who provide simultaneous multi-party guidance through speech-to-text communication, annotated screenshots, and document display on the glasses. The system is designed to automatically record video, audio, chat, and operation logs from every session; these records will be processed using AI for key-scene extraction, transcription, and work-process summarization, then restructured into reusable training content with planned integration into the Moodle learning management system. Rather than treating remote support as a temporary communication function, TSM captures the judgment process of experienced personnel, what they observe, prioritize, and decide, and converts it into organizational knowledge. A pre-implementation survey of 146 respondents found 80.6% willing to use the system and 85.2% supporting the initiative. A pilot deployment with ten smart glasses units is underway: the time less-experienced staff can operate independently under remote support has tripled, and re-inspection rates have decreased by 50%. TSM is designed as a replicable model for small municipalities nationwide, with potential applications in construction, agriculture, and disaster response.
INTRODUCTION
Municipal water and sewerage services in Japan are maintained by a shrinking technical workforce: veteran engineers who can judge the condition of pumps, valves, and pipelines from sound, vibration, and visual cues are retiring while the facilities they maintain grow older. When the few experienced individuals in a small municipality are unavailable, inspection quality and emergency response both suffer.
Remote support tools and training systems each address part of this problem, but separately: video calls connect a field worker to a remote expert, yet the exchange disappears once the call ends, while training systems depend on someone authoring content after the fact, which rarely happens in understaffed utilities.
This paper presents Type Soo Mine (TSM), a browser-based smart glasses platform jointly developed by Soo City, Kagoshima Prefecture, and a private XR development company, positioned as Japan's first browser-based infrastructure digital transformation (DX) model integrating field sharing, recording, and education in one workflow. Unlike systems designed primarily for skills training, TSM starts from remote work support in actual operations and treats every support session as a recording to be reused.
We address the following research question: how can a browser-based smart glasses platform transform the real-time remote support that small municipal utilities must perform daily into structured, reusable organizational knowledge and training content, without requiring separate documentation effort from already understaffed field teams?
RELATED WORK AND POSITIONING
Smart glasses have been applied to remote assistance in manufacturing, plant maintenance, and healthcare, typically through dedicated applications tied to specific devices; for small municipalities with limited IT staff and mixed device environments, such installation and account management are costly. A second gap concerns what happens after the call: prior deployments generally treat the session as transient communication, and few mechanisms capture veteran judgment during routine work before it is lost at retirement. TSM is positioned at this junction: remote support as the capture point, training content as the downstream product.
SYSTEM DESIGN
Architecture. TSM is dedicated-application-independent: field workers wear smart glasses that stream first-person video and audio through a browser-based platform, and remote participants join from personal computers, smartphones, or tablets without installing software, reflecting the reality of small municipalities and partners whose device environments cannot be standardized.
Multi-party remote support. A session connects the field worker simultaneously with remote supervisors, equipment manufacturers, and maintenance contractors; guidance flows through speech-to-text communication, annotated screenshots, and document display on the glasses, so a worker in a noisy pump station can receive instructions visually, hands-free.
Recording and knowledge conversion. The system is designed to automatically record video, audio, chat, and operation logs from every session, with the intent of capturing the judgment process of experienced personnel, what they observe, prioritize, and decide, rather than only the procedures they perform.
AI processing pipeline. The AI processing pipeline is currently in the design phase and follows three planned stages. First, automatic speech recognition converts session audio into searchable transcripts. Second, a scene-segmentation step identifies candidate key moments in the first-person video, using cues such as annotation events and dialogue activity rather than manual review of full recordings. Third, transcripts and identified scenes are summarized into structured work-process descriptions, from which draft captions and assessment questions are generated. Because these outputs inform field safety judgment, all AI-generated content is intended to be reviewed by a domain expert before publication to learners through the Moodle learning management system.
DEPLOYMENT
Before implementation, we surveyed 146 respondents connected to the city's water and sewerage operations; 80.6% expressed willingness to use the system and 85.2% supported the initiative, indicating that field acceptance was not a major obstacle.
A pilot deployment using ten smart glasses units is underway in water and sewerage inspection operations in Soo City; veteran staff and external specialists support less-experienced workers remotely, and first-person records are already being captured digitally and reused within the organization.
PRELIMINARY RESULTS
Two preliminary indicators are reported from the ongoing pilot. First, the time that less-experienced staff can operate independently under remote support has roughly tripled compared with prior practice. Second, re-inspection rates have decreased by about 50%: problems are more often resolved correctly on the first visit when remote specialists can see what the field worker sees.
We treat these figures with caution, as they come from a single municipality, a limited number of units, and operations still adjusting to the platform. Quantitative evaluation of latency, first-response accuracy, and training-material effectiveness is in progress and will be addressed in future work.
DISCUSSION
We draw three initial observations from the pilot. First, browser-based architecture mattered: external manufacturers and contractors joined sessions without installation steps, shortening the path from problem found to specialist consulted. Second, positioning recording as a by-product of support work avoided the authoring bottleneck that stalls many knowledge-management efforts; records accumulate because the work itself requires the session. Third, field staff accept being recorded when the immediate benefit, help while working, is concrete.
We also acknowledge several limitations: the evidence base is one municipality and ten units; the AI pipeline and Moodle integration are designed but not yet operating at scale; the two headline indicators lack controlled baselines; and environmental conditions such as humidity, lighting, and network coverage have not yet been analyzed systematically.
Further work will expand data collection across sites, personnel, and operational conditions to examine generalizability, and will explore transfer to adjacent fields such as construction, agriculture, and disaster response.
CONCLUSION
We presented TSM, a browser-based smart glasses platform linking remote work support, session recording, and training-content generation in a single workflow. A pilot with ten units in Soo City indicates that less-experienced staff can work independently roughly three times longer under remote support and that re-inspection rates have fallen by about half, while first-person records accumulate for reuse. These are early results and broader validation is in progress; even so, for small utilities facing staff decline, remote support and knowledge preservation need not be separate investments, since a browser-based platform can serve both from the same daily work.