In manufacturing, 76% of job candidates noted that safety in the workplace was either very important or important, ranking higher than workplace flexibility and even pay and benefits. As a continual focus, manufacturing leaders have seen advancements in safety lead to greater engagement and trust with employees—driving greater success in the long run. And according to a recent Bureau of Labor Statistics report, nonfatal injuries in the manufacturing sector decreased by 21,400, a 6.15% reduction, to 326,400 cases in 2023.
As artificial intelligence (AI) spreads across workplace domains, manufacturers are exploring opportunities to implement AI within safety operations. Organizations have long used safety walks, hazard identification tools, safety suggestions, and near-miss reports, all manual and user-identification heavy. In Manufacturers Alliance’s recent member survey, 72% of respondents believe AI use will make plants safer, and most believe that training and simulation safety protocols will be the top area that AI can impact, followed by tracking safety KPIs, both ripe with the ability to transition out of manual processes.
Will AI Make Manufacturing Plants Safer?
Source: Manufacturers Alliance member survey, 2024.
Training and Tracking
Manufacturers routinely utilize a variation of platforms and learning styles to accommodate the needs of their people and operations. Seventy-seven percent of survey respondents believe technology can improve training and simulation safety protocols. Realistic training modules and simulations allow workers to practice safety procedures routinely to keep skills sharp.
Manufacturers continually face age gaps between long-tenured employees and other generations, and the comfort levels with technology often follow those lines as well. “Younger generations are used to swiping strokes on an iPad to make things happen, and many of our manufacturing systems don’t respond to that type of interaction,” added one manufacturing executive, but the gap is getting smaller. Jim Beilstein, Vice President of Global Operations and Supply Chain, Insulation, at Owens Corning, shared how they “are looking at a whole connected worker program as an opportunity to increase the speed to proficiency of employees, especially given that we’re hiring new people on a regular basis,” blending the use of technology to overcome technological knowledge gaps. When employees are more comfortable and knowledgeable, operations become more stable and safer.
Immersive training environments like augmented reality and virtual reality (AR/VR) can allow for controlled, hands-on training without the risks associated with a live environment. Gamified learning experiences bring a greater amount of engagement and interaction, aiming to adapt to the learning types of different employees and providing stronger training to avoid later safety risks. AI providers have recognized that AI-based training helps standardize work while also providing a smooth training experience that targets continuous improvement. One industrial manufacturer partnered with providers to fully replicate their exact equipment and create a VR experience with real controls. Participants receive a score on both virtual and physical movement as they operate through the virtual yet extremely realistic courses.
Owens Corning uses technology to offer blended training methodologies to which newer employees are more accustomed. Closed captioning and a “YouTube style” narration allows for employees to access video training live while safely completing a job. “Instead of having to sit in the classroom for three hours and then go on the floor with somebody, they can simultaneously learn and act. Incorporating technology into training has a lot of potential in the power for driving speed to efficiency,” shared Jim Beilstein.
Similarly, they are in the early phases of working with technology to assess predictive analytics, tracking metrics around the facility to predict whether the risk level for potential recordable injury is increasing or decreasing. Their metrics are “safety, operations, and human related,” added Beilstein, as they factor in numbers like days a plant doesn't have an injury, turnover, and overtime. With targets across many of their facilities, Owens Corning is set to standardize risk measurements and goals aimed at risk reduction every single year. Eighty-two percent of survey respondents are already analyzing data on near-miss reports, but only 26% are using predictive analytics for risk assessment.
Their programs even outstretch to condition-based maintenance so they can “get ahead of abnormal situations.” Preventive maintenance keeps consistency, but continual monitoring decreases the likelihood of a surprise breakdown. While not designed as a safety tool, there’s clear safety implications that go together with the reasoning for instituting that technology and considering the overall productivity and safety of their facilities.
Physical Monitoring
Employee health and safety monitoring is the third most popular target that manufacturers expect technology to improve (59%). High-accuracy computer vision models can detect unsafe acts and conditions in real time to immediately alert employees to correct any behaviors that would expose them to unsafe situations. This can include monitoring the employee for wearable safety (i.e., ensuring that they have safety glasses or proper PPE) or correcting body mechanics, such as adjusting lifting or twisting motions. “We have to first resolve how good are the predictions and how good are the models—and train them all in the right way,” said Beilstein.
Audi has been implementing ergonomic exoskeletons to prevent excessive physical strain on employees. The right technology can take over some autonomous and physically straining activities, removing the wear and tear on employees and allowing for their skills to be used elsewhere.
For those with fewer repetitive motion-based injuries like at Owens Corning, exoskeletons were not the right answer. They’ve set a rule that employees don’t lift anything over 40 lbs. alone. Instead, they use a winch or other tools to assist or two employees lift together. They mostly find the issue with maintenance and repairs. “It’s people that are picking up small pumps or similar items and moving them,” noted Beilstein. “We’ve tried exoskeletons, and in our operations we found it doesn’t de-risk the process in a way that we’d like to see.”
Article Continues
Tony Pagliaro, Director of Operations at Tennant Company, has video monitoring in place to map out any situations where an incident may have occurred, pinpointing a single point in time that they can then review with the appropriate staff to address any needed process or behavioral changes. With ergonomic injuries, employees can then take robust action items to prevent any further injuries and work with their on-site physical therapist. Although reactive, the process of reviewing incidents and addressing issue points is set to change future behavior and decrease injury repetition.
AI video monitoring is aiding a an industrial manufacturer in their supervision and PPE compliance. They began their implementation by using cameras to detect movement into “red zones,” or areas that are not safe for employees. If cameras detect someone entering a red zone, an email is automatically sent to supervisors. The organization has continued their programming to also include proper PPE identification, which took more refinement. The safety leader has seen improvements in supervisors’ accountability to take corrective action and ensure best practices in areas where there are greater safety risks.
Many manufacturers have reported accidents involving powered industrial vehicles (PIVs), and this is a focus area for preventive measures. At Owens Corning, cameras on their lift trucks detect and predict when employees are within striking distance and automatically reduce speed or shut down. “It’s a little different from the warning lights and other indicators that are typically used, but we are continuing to investigate and experiment with more,” noted Beilstein.
Additionally, many manufacturers have incorporated real-time environmental condition monitoring, such as temperature and air quality, to enhance conditions in the workplace, which most of our respondents indicated they were doing. In particular locations or in certain industries, keeping employees comfortable and safe can be a competitive advantage.
Overcoming Challenges
Implementing AI into safety is newer, and therefore it is harder to prove ROI. At 66%, the top concern for survey respondents was cost. Implementing AI can initially incur high capital expenditures compared to the long-term cost benefits. To combat initial costs, companies are targeting technology to areas with higher safety risks, and the gains made from reducing accidents provide subsequent cost reductions.
Challenges or Concerns with Implementing AI for Safety in Manufacturing Plants
Source: Manufacturers Alliance member survey, 2024.
Yet, the biggest challenge we heard from interviewees about AI implementation for safety involved employee acceptance, which over half of survey respondents were concerned about. Pagliaro shared how internally at Tennant they would partner with human resources, legal, and EH&S to ensure compliance with their safety measures and collectively ensure that they are being transparent about the usage of the tracking and what that means for employees.
“There’s a balance between taking the next step in your safety journey [with video tracking] and ensuring that feedback from the videos is not used as discipline or negative consequences on the employee,” said Pagliaro. There would still need to be “trust in the leadership to not say ‘Hey, we noticed on camera you were three minutes late coming back from break,’ as that is not the intent.”
Employee acceptance of new technology and data security and privacy were tied for the third concern for survey respondents. We heard from interviewees who are blurring faces and creating strict video archiving guidelines, ensuring limits on what the videos can be used for. Some manufacturers mentioned partnering with other departments to ensure that proper rules are in place to maintain data security, and ultimately privacy, including legal, HR, and IT.
At the end of the day, safe operations depend on the employees keeping safety at the top of mind, and it’s important to not let tenure and comfortability rule. “We’ve seen 20–30-year employees getting hurt, especially in maintenance, because they have done the task hundreds if not a thousand times. There is just something slightly different one time that they were not anticipating, and they did not do enough of a pre-job hazard analysis to recognize it ahead of time,” said a senior safety executive. Safety and technological advancements and measures can go a long way, but they cannot replace human process and conscious actions.
“At the end of every day, with every process that we have, every injury, we’re looking for how we can better engineer processes to reduce risk,” noted a safety leader for an industrial manufacturer.
Want More?
Members have full access to the complete results, Safety in Manufacturing: Technology Use.
