Bottom Line: AI and machine learning are reducing construction site accidents, theft, vandalism and hazardous operating conditions by analyzing 24/7 video feeds in real-time, gaining new predictive insights and contextual intelligence into threats.
According to the National Equipment Register, construction theft losses often exceed $1B a year. The latest model equipment, tools and supplies are the most stolen and the least likely to recover. Only 25% of stolen construction equipment is recovered. Add to that an estimated $13B annually that the cost of accidents and injuries cost the construction industry every year. The urgent need to improve construction site security becomes clear.
Improving Construction Site Security With Machine Learning
Analyzing how the specific conditions, factors, locations and phase a given construction site is operating in contribute to greater site security and safety risks is a perfect application of machine learning. Today and in the future, remote monitoring systems in use to protect construction sites will rely on supervised machine learning algorithms to discover new patterns in historical data. Machine learning-based remote monitoring systems rely on IoT sensors combined with night vision, infrared and thermal sensing cameras to capture real-time data streams.
By combining historical video feeds and images with real-time data feeds, machine learning-based remote monitoring systems provide predictive insights into when potential accidents, thefts, or hazardous operating conditions could occur. AI-based remote monitoring systems providers are setting a fast pace of innovation in this area, which is also reflected in the intuitive design of their dashboards and platform approach to scaling these systems globally. Leaders in this area include Twenty20 Solutions, whose cloud-based platform and approach to usability set the standard in remote construction security. Twenty20 Solutions’ remote monitoring system is entirely browser-based, supports geo-location applications (e.g., RFID, GPS and radar asset tracking) and has a user-customizable dashboard. The following is an example of the Twenty20 Solutions dashboard:
10 Ways AI Is Improving Construction Site Security
One of the most valuable takeaways from a recent conference call with construction safety and security leaders who have standardized machine learning-based remote monitoring systems is how much time they save from false alarms. A security director managing construction projects in progress in Miami, Atlanta and Chicago says machine learning has virtually eliminated false alarms at his construction sites. “Our team has fine-tuned machine learning algorithms to the specific patterns of our operations and it’s virtually eliminated false alarms – and focused on predicting theft and break-in attempts much more accurately,” he said.
Based on insights gained from the conference call, here are 10 ways AI is improving construction site security:
1. Reducing the dependence on on-site security teams alone and gaining a 24/7, 365-day monitoring view of each construction site. All construction safety and security leaders say this is the most valuable aspect of a cloud-based remote monitoring system capable of accepting data from IoT, digital, thermal and infrared cameras. Having teams on the ground are still essential, yet having the real-time, always-on monitoring providing a data stream to train models is invaluable.
2. It is possible to reduce workplace injuries and potential liability litigation by identifying which workers are wearing Personal Protective Equipment (PPE) or not. Using advanced pattern matching supported by supervised machine learning algorithms, construction safety and security leaders can identify when workers in high-risk roles and work zones are wearing PPE or not. Construction sites continue operating during Covid-19 and in many instances, everyone needs to have a face shield and masks on for the site to stay in compliance with CDC guidelines. Remote monitoring systems can tell immediately which work teams need coaching to remain in compliance.
3. Replace checklists, routines and other manually based approaches to auditing security with real-time monitoring capable of providing trending and image analysis if needed in seconds. Safety and security leaders say their biggest challenge today is saving the valuable time of construction teams on-site. They don’t know when a lockdown or quarantine will be reintroduced in the cities they have projects in. According to leaders spoken with, combining machine learning techniques with site data always accessible on an intuitive dashboard reduces manual workflows by a factor of 5 or more.
4. The better the construction site’s real-time safety and security monitoring, the more flexible construction processes become. Safety and security leaders say that the better a construction site gets at capturing and interpreting real-time sensor and video data, the more adaptive their construction processes become. What initially starts as an investment to achieve safety and security goals turns into process improvement gains based on new data insights.
5. Advanced video analytics that supports smart tags and contextual cues of key events are coming, which will further improve the predictive accuracy of machine learning-based remote monitoring systems. Efforts to gain greater contextual intelligence from every video frame are driving R&D spending across the remote monitoring industry. Expect to see patents that define the future of advanced pattern matching and predictive accuracy of remote monitoring systems next year.
6. Predict the impact of combined cyber and physical threats on construction projects in real-time, creating specific algorithms to assign risk scores to a particular series of events. What’s most fascinating about the ongoing development work in remote monitoring systems is how physical and cyber events can be correlated. For example, suppose there’s a cyberattack on a specific construction site’s systems. In that case, it’s possible to increase the monitoring and predictive accuracy of which physical systems attackers are targeting, thwarting thefts, breaches and destruction project itself.
7. Assigning risk scores to each construction project then analyzes which factors need to be improved to reduce equipment and material theft threats. Building predictive models using machine learning based on theft-related data can help predict when another attempt will occur. Safety and security leaders say that’s why having video data from the very beginning of each project is so valuable; it can be used to train machine learning models and predict another theft attempt.
8. Reduce the threat of industrial espionage by identifying and tracking intruders’ activities on a job site who don’t have access credentials. Industrial espionage in construction can lead to billions of dollars of losses, especially for high tech manufacturers building chip foundries that include proprietary technologies. Keeping sites secure needs to start with an effective AI-based remote monitoring strategy.
9. Discovering where, when and how pilferage of construction supplies is happening, so costs are kept in check and worksites made more secure. From 10 to 25% of supplies on large-scale construction projects are stolen and resold every year. Stopping pilferage means the difference between a construction project coming in under budget or not, which is why safety and security leaders are turning to remote monitoring to get this problem under control.
10. Ensuring construction sites stay in compliance with OSHA and related government compliance requirements while at the same time creating a real-time audit record for government auditors. Safety and security leaders say that preparing for an OSHA audit can take weeks or even months of advance prep work. Real-time monitoring systems can produce reports on demand, saving hundreds of hours a year in audit prep time for government agencies.