By EMA Manager of Employment Relations and Safety Paul Jarvie
Manufacturing continues to sit among New Zealand’s highest-risk sectors for workplace injury. Forklifts, heavy machinery, manual handling and complex site layouts all create situations where the margin for error is small. While most manufacturers invest heavily in controls, training and systems, too often the first clear signal that something isn’t working is when someone gets hurt.
Emerging technology is beginning to shift that equation, from reacting to incidents to identifying risk patterns before harm occurs. One of the more promising developments comes from RUSH, a New Zealand technology company applying artificial intelligence (AI) to existing CCTV systems. The goal isn’t to surveil workers, but to give businesses better visibility of everyday risk using data they already have.
This work is underpinned by R/VISION, an AI-powered computer vision platform developed by RUSH. R/VISION provides near real-time insight into workplace operations, using existing camera infrastructure to help businesses identify risk, strengthen safety practices and optimise operational efficiency across New Zealand manufacturing sites.
As Caleb Perelini from RUSH explains, their approach is about adding ‘another set of eyes’ to the workplace. “We’re not installing new cameras or creating new footage,” he says. “All we’re really doing is taking cameras that already exist on site and using AI to look for the same things a health and safety professional would be worried about.”
Those things include people getting too close to moving plant, forklifts exceeding safe speeds, pedestrians entering exclusion zones, or PPE not being worn correctly. Increasingly, the technology is also being used to identify high-risk movements such as unsafe lifting postures that can lead to serious musculoskeletal injuries over time.
Many of these risks do not easily show up in traditional reporting systems. Near misses, unsafe habits and gradual ergonomic strain rarely appear in incident registers, yet they account for a large proportion of long-term injury costs in manufacturing.
“What AI allows us to do is build a quantitative picture of risk,” Caleb says. “Instead of guessing how risky a workplace might be, you can actually see how often certain things are happening, where they’re happening, and whether the changes you’ve made are working.”
The data is quantitative, not assumed or relying on nil incident reports. That information is typically used to improve systems rather than focus on individuals – ‘fix the work, not the worker’. Several manufacturers have used insights from this data to redesign traffic management plans, adjust site layouts, or change workflows that were creating unnecessary exposure to harm.
In one example, analysis showed workers coming within half a metre of moving forklifts up to 100 times in a single shift. A simple layout change dramatically reduced those interactions. “Within a couple of weeks, they saw reductions of around 60 to 80 per cent in those high-risk events,” Caleb says, with improvements sustained over time.
Understandably, privacy concerns often arise early. Workers and employers alike want reassurance that technology is not becoming surveillance. Caleb is clear that identifying individuals is not the intent or the design. “We don’t identify individuals, we don’t do facial recognition, and we don’t build the system to be disciplinary,” he says. “The focus is on trends, how often something is happening and why, because that usually points to a policy, design or training issue.”
This distinction is important. When technology is framed as part of a learning and prevention system rather than an enforcement tool, it is far easier to earn trust and engagement across the workforce.
This work is supported by ACC through injury prevention funding with a focus on reducing time-lost injuries in manufacturing. Participation does not involve sharing data with regulators or insurers. “Nothing is automatically shared with ACC, WorkSafe New Zealand or anyone else,” Caleb says. “ACC’s interest is whether technology like this reduces long-term injury costs, not enforcement.”
From the EMA’s perspective, the value of this technology lies in its ability to support better conversations about risk. It moves businesses beyond lag indicators such as injury statistics and toward leading indicators that enable trends to be analysed and measure interventions that have been trialled, where harm could occur.
AI is not a silver bullet. It does not replace leadership, worker engagement or good system design. But used well, it can strengthen all three by making risk more visible and decisions more informed. For manufacturers focused on preventing harm rather than reacting to it, that is a powerful step forward.
Click here to find out more about R/VISION technology.