Why robot safety must evolve as machines become smarter and more autonomous

By Jenny Shern, general manager at NexCOBOT, part of Nexcom

Robot installations worldwide have more than doubled in the past decade – and deployments are no longer contained to the manufacturing line.

A growing wave of players are racing to develop advanced robots and humanoids with sophisticated capabilities in a variety of dynamic environments, from retail stores and hotel lobbies to eldercare facilities and bustling warehouses

For instance, Amazon’s newest collaborative robot, Vulcan, is equipped with tactile sensing that enables it to pick, sort, and stow objects with the finesse of a human hand.

These advancements signal a not-so-distant future where human-robot collaboration becomes a seamless part of everyday life.

But as we scale the heights of what robots can do, we must also rethink how we ensure they do it safely. If we want robotics to perform effectively in unpredictable, human-centric environments, we must treat safety as a foundational principle.

Safety shouldn’t be retrofitted. It should be the framework that you build everything else upon.

Whether through tightly integrated safety units or modular add-ons, systems should not just detect danger, but actively define how robots move, interact, and adapt in real time.

Safety as a core design framework

As robots become more dexterous and perceptive, their ability to interact with unstructured, real-world settings has dramatically improved.

Vulcan’s tactile feedback system, for instance, lets it handle awkwardly shaped or dimly lit objects in ways that would challenge traditional vision-only robots.

However, there are still major questions about how safely robots can adapt to the dynamic environments in which they now operate.

Environments like warehouses, hospitals, and homes are inherently unpredictable and potentially dangerous, filled with changing light conditions, unexpected obstacles, and most importantly, vulnerable humans.

Protections that would be useful on a factory floor, like pre-programmed motion paths and physical safety cages, are no longer sufficient.

Without a robust safety infrastructure, advanced robotic capabilities meant to boost efficiency can actually magnify the risk of failure and ultimately harm.

In fact, the Center for Investigative Reporting found that Amazon warehouses with robots experienced 50 percent more serious injuries.

What we’re learning is that advanced robotics requires a layered, multi-domain approach to protection. Internally, robots must be equipped with dedicated safety controllers that regulate joint motion, velocity, and force.

Externally, robots need intelligent sensing systems to understand and predict human behavior, estimate future motion, and adjust their own trajectory accordingly.

These capabilities demand significant planning, rigorous testing, and a long-term commitment to system integration, yet many developers are skipping these steps in favor of faster release cycles.

Safety is too often treated as an afterthought or sidelined for other priorities. In the push to be first to market, some developers are putting speed over structure, rushing out impressive prototypes that lack the foundational safety systems required for real-world deployment.

Integrators are left scrambling to bolt on sensors, emergency stops, or perception modules after the prototype is already baked, while developers are under pressure to churn out new robot form factors and use cases that prioritize proof-of-concept over platform stability.

It’s a dangerous pattern. Enhanced capabilities introduce greater complexity – and with it, greater consequences when something goes wrong. The smarter, stronger, and more autonomous a robot becomes, the more impactful its mistakes can be.

5 foundations for functional safety

Functional safety can’t be a checkbox to tick at the end of development; it needs to be a design philosophy that informs every choice from concept to deployment.

To help you put that philosophy into practice, here are five key considerations to guide your approach.

1. Integrate safety from the start

Scaling robots from design to production to real-world deployment requires building safety into every component, from motion planning and AI decision-making to environmental perception and control logic.

Delaying that work can lead to expensive redesigns and integration bottlenecks down the line.

The responsibility for building safety-ready robots must shift upstream, from integrators and developers to those designing concepts and building system architecture.

From Day One, take a systems-level mindset where safety is woven across disciplines and build it into your architecture and design choices.

For instance, the design of core safety protocols should account for things like passive and active fall strategies – how the robot should respond if it loses power mid-step – and context-sensitive behaviors, like slowing down in narrow corridors or yielding to a person carrying an object.

2. Scale safety with a modular approach

Designing with a modular architecture enables safety functions to grow in lockstep with your robot’s evolving capabilities.

By separating safety algorithms, vision systems, and fallback mechanisms, you can ensure that each component scales independently while maintaining system-wide integrity.

Modularity also brings flexibility. It allows you to upgrade or replace components without reengineering your entire platform, and lets you respond quickly to new safety regulations or customer requirements.

3. Build on open architecture

Robotic systems must juggle real-time decision-making, AI inference, and complex motion control – often running simultaneously on a shared hardware platform.

Meeting those demands requires substantial computing power and smart architectural design.

An open architecture provides the flexibility to integrate specialized safety modules or hybrid controllers that coordinate motion, perception, and control systems in real time.

Leveraging open standards like EtherCAT or ROS 2 expands your access to a broad ecosystem of compatible sensors, drives, and safety components — making your robots easier to maintain, upgrade, and scale for deployment in new environments.

4. Consolidate workloads strategically

Due to space constraints and other limitations, robotics designers often consolidate systems that are traditionally separate.

For example, motion control and AI perception – once run on separate hardware – are now frequently combined onto a single platform to reduce wiring complexity and minimize footprint in compact systems like humanoids.

This approach enables more efficient distribution of compute workloads across the system, which creates an opportunity to integrate safety more deeply.

Increasingly, developers are turning to layered safety frameworks and hybrid architectures that ensure safety is a central orchestrator that actively shapes how the robot thinks, moves, and responds.

As you design your architecture, consider carefully which functions benefit from shared hardware, such as motion control and AI inference, and which are best kept separate, like critical safety protocols that demand independent fail-safes.

This approach enables more efficient distribution of compute workloads across the system, while ensuring that safety acts as a central orchestrator rather than a reactive backup.

5. Go slow to go fast

Prioritizing speed over safety is a costly (and potentially dangerous) trade-off. Without robust functional safety, early wins can quickly unravel when faced with dynamic environments, certification hurdles, or integration complexity.

As a result, taking the time to embed layered safety frameworks and hybrid control architectures from the beginning is often a smarter acceleration strategy.

When safety is woven into how the robot thinks, moves, and responds – not just how it stops – you gain a platform that’s built to adapt, scale, and succeed far beyond the prototype phase.

Designing for innovation that will last

As robotics moves from controlled settings into the chaos of real-world, human-centric environments, functional safety will be critical to prove that these systems can operate reliably, intelligently, and responsibly alongside people.

That trust won’t be earned through flashy demos or rapid releases, but rather through thoughtful architecture and safety-first design.

Because in the end, safety isn’t what slows innovation down; it’s what makes innovation durable.

About the author: Jenny Searn is skilled in automation IoT, robotics and AI retail business development, sales management, strategic partnerships, and product and project management. Strong business professional in building companies and team management.