The municipality of Edmonton has carried out periodic reviews of the programs and services that the city has been providing since 2016. These reviews involved the employment of an internal team to evaluate the city’s services with an impartial eye and ensure that the strategic objectives serve individuals as needed and save costs. As part of these efforts, the Canadian city of Edmonton offers Safety Codes Inspection Services (also known as building permits) in order to monitor commercial, industrial, institutional, and residential development activities to ensure the safety of residents. To guarantee the safety of buildings in both small and large projects, the municipality collects application fees, examines plans, issues permits, and conducts inspections.
In light of the increasing number of inspections and the overburdening of employees and municipal resources, the 2019 internal team report recommended harnessing AI technologies to come up with a predictive model for building inspections and stop wasting the city’s resources by limiting inspections to higher-risk sites. This model aims at assessing construction compliance with applicable safety codes without the need for inspections at low-risk sites. It will direct the municipality’s resources towards more complex inspections to avoid obstructing the progress of projects. This model also determines the best way to leverage internal and external inspectors based on the previous and expected workload, thus allowing greater flexibility in the appointment process.
The municipality of Edmonton has collaborated with the University of Alberta to launch the predictive model that relies on several indicators, including previous and current data to predict future results. This AI-based model can forecast whether contractors would grant approval for low-risk inspections, thus allowing the city to automatically approve inspections with unprecedented results in terms of abiding by safety codes in buildings. On another note, the model reduces unnecessary inspections for experienced employees and directs resources toward advanced inspections that pose a threat to public safety.
The municipality has conducted building safety inspections for 600,000 detached households for over ten years. These visits generated a large amount of useful data, including information about contractors, geographic location, building information, and previous inspections.
The predictive model is based on a mechanism and sequential stages. First, detached homes are prepared to undergo a safety code evaluation through the artificial intelligence program to determine if inspections are necessary. Second, the model predicts the success or failure of the plumbing stack, plumbing groundworks, HVAC stack, and HVAC concealed duct inspections. Finally, contractors who have requested inspections are notified by e-mail at 10:00 am the following day to inform them whether they are dealing with high- or low-risk sites.
The model evaluates multiple attributes to determine if the building needs an inspection or not. It begins by determining the attributes of the building, including permit degree, built-up area, building design, the location of the vacant building space, the parking space, and the building value. The model then considers the geographical information that includes the classification of the neighborhood and then attributes about the contractor, in addition to the number of inspections conducted over 12 months and the admission rate from the first time till the last 12 months. Finally, the model evaluates the attributes of the inspections, including inspection type and date, and the interval between applying for the permit and conducting the inspection.
The process of approving construction begins when building permit applications are received and fees are collected, after which the building plan is studied and reviewed. Then, a permit for construction and underground electrical installations is granted. After completing all these steps, the contractor can start construction and begin inspections until the occupation permit is issued, thus allowing the house to be ready for accommodation.
To involve as many entities as possible in the development of this model, the municipality consulted with a number of internal and external partners to listen to their opinions and suggestions before the official launch of the AI-based inspection mechanism. Researchers from the University of Alberta participated in the technology conference in Chicago and presented their feedback and proposals on the implementation of the project. This model has proven to be effective in terms of inspections to assess safety codes as it has helped reduce the time it takes for contractors to obtain approval, avoid backlogs during peak times, and reduce the number of non-mandatory low-risk inspections. This helps residents move faster into their homes, in addition to stimulating economic development and boosting the local economy.
Testament to the project’s success, the city was awarded the Smart 50 award in the digital transformation category by Smart Cities Connect. The Smart 50 Awards recognize cities that successfully implemented projects with a significant impact at the municipal scale. This award was well-deserved, especially since this model succeeded in reducing the number of qualified inspections by 37% since its launch in October 2019. Annual inspections are expected to drop by a third.
