Survey Exposes Software Architecture as the Critical Bottleneck in Robotics Development

QNX revealed findings from its "Inside the Robot: Architecture Benchmark Report" at the Robotics Summit & Expo this week, and the data should make engineering teams rethink where they're spending their optimization efforts.

The study, conducted by OnePoll across 1,000 software developers and engineers working in robotics, found that 27% of respondents identified software architecture and integration as their primary performance bottleneck. That's nearly double the 16% still blaming hardware. I find this gap telling: we have collectively spent years throwing better processors and sensors at robotic systems while the software foundation quietly became the actual constraint.

The implications extend beyond development timelines. As robots move from factories and warehouses into city streets, hospital wards, and surgical suites, the stakes for software reliability have fundamentally changed. More than four in five respondents (83%) reported their systems now operate alongside humans. Among organizations that haven't yet deployed in human environments, two-thirds (67%) expect to within three to five years.

The Real-Time Problem Nobody Wants to Talk About

Here's where I think the industry has a credibility gap. Nearly all respondents (95%) stated that deterministic, real-time execution matters to their systems. Yet 91% run these workloads, at least partially, on general-purpose operating systems. When asked why they stick with GPOS despite the mismatch, respondents cited legacy codebases, team familiarity, and tooling ecosystems. Those are honest answers, but they don't change the fact that you cannot reliably certify safety-critical behavior on an OS designed for throughput over determinism.

The market appears to recognize this contradiction. Eighty-six percent of GPOS users said they are open to changing their operating system. QNX noted that safety-certified commercial solutions consistently ranked as the best fit for their actual requirements, not what they're currently using.

Where Investment Is Actually Flowing

Looking ahead, 85% of developers expect software to play an even greater role in robotics over the next three to five years. Their planned investments tell the story: AI-driven decision-making (51%) and cybersecurity (51%) rank as top priorities, followed by operating systems and real-time control software (37%).

I notice the OS and real-time control category trails AI and security in stated priority, which seems disconnected from the earlier finding that architecture remains the dominant bottleneck. Perhaps teams view AI and security as new workstreams, while OS infrastructure feels like a solved problem they can address later. That would be a mistake.

"Developers consistently cite four core challenges: integration complexity, certification delays, functional safety risks in human-machine interaction, and ensuring predictable behavior when it matters most," said Jim Hirsch, global vice president of sales and general embedded markets at QNX.

Those challenges won't disappear by upgrading sensor packages or adding more compute. The report suggests the industry knows this but hasn't fully acted on the knowledge.

What This Means for Engineering Teams

The data makes the path forward relatively clear: software foundations are now strategic assets. Teams treating architecture as secondary to algorithm development will continue hitting the same walls. The question isn't whether to invest in better software infrastructure. It's whether you can afford to keep deferring it while competitors solve the same problems faster.

M4S TAKE

My take: AI claims need scrutiny. The useful implementations reduce cycle time or defect rates in measurable ways. Vague promises about 'optimization' without specific metrics are usually marketing.

Simon McLoughlin