Industry Standards

Safety, ethics, and best practices for humanoid robots at work and at home.

Industry Standards & Best Practices: Making Humanoid Robots Safe

Picture walking into your workplace tomorrow and meeting your new colleague – a 6-foot-tall humanoid robot that walks, talks, and works alongside you. It sounds like science fiction, but it's happening right now. Companies like Amazon, BMW, and Tesla are already testing humanoid robots in their facilities. But here's the critical question: How do we make sure these sophisticated machines don't accidentally hurt anyone?

The answer lies in a complex web of safety standards, ethical guidelines, and best practices that engineers, lawmakers, and safety experts worldwide are racing to develop. As humanoid robots transition from movie screens to factory floors – and eventually to our homes – creating robust safety rules has become one of the most important challenges of our time.

Why Humanoid Robots Need Special Rules

Traditional industrial robots have operated safely in factories around the world for decades, but they've always been kept behind protective barriers – metal cages, light curtains, and safety zones that immediately shut down the robot if a human gets too close. This approach worked perfectly for traditional robotic arms that stayed in one place and performed repetitive tasks. However, the discussion about why existing safety measures aren't enough for humanoid robots continues in The Surprising Truth About AI in Robotics Nobody Tells You, where experts examine how AI allows robots to make decisions and adapt to new environments autonomously.

Humanoid robots change everything. These robots pose unique safety risks that are driving a push for new standards before they start sharing our workplaces and homes. Unlike their stationary cousins, humanoid robots are designed to walk around, navigate complex environments, and work directly alongside humans without protective barriers.

Here's the fundamental challenge: when a traditional industrial robot loses power, it simply stops moving – this is called "failing to safe." But when a humanoid robot loses power, it falls down, potentially creating an entirely new hazard. This basic difference requires completely rethinking how we approach robot safety. Research has shown that most robot-related injuries happen during maintenance and repair work, not during normal operations.

The Global Push for New Safety Standards

International Standards Take Center Stage

The foundation of robot safety worldwide rests on ISO 10218, the international standard for industrial robot safety. After nearly eight years of work, this standard received a major overhaul in 2025, marking the first significant revision since 2011. The updated standard is much more comprehensive – some sections are now twice as long as the previous version, reflecting how much robotics technology has advanced.

But ISO 10218 was designed for traditional factory robots, not for humanoid machines that walk among us. That's why experts are developing ISO 25785-1, a groundbreaking new standard specifically for legged robots and advanced mobile robots. This standard addresses the unique challenges of dynamic stability, fall prevention, and human-robot interaction that traditional robot safety standards simply can't handle.

The development of this revolutionary standard is explored in detail in New Robot Safety Standards Explained: Insights from A3, Universal, where leaders from Agility Robotics, Boston Dynamics, and Universal Robots discuss how the proposed ISO 25785-1 aims to close safety gaps for mobile robots with actively controlled stability.

The leadership group of the U.S. delegation working on the ISO 25785-1 standard consists of Federico Vicentini from Boston Dynamics, Kevin Reese from Agility Robotics as Project Leader, and Carole Franklin from the Association for Advancing Automation (A3). This Type C safety standard will provide much needed clarity and requirements for humanoid robots with actively controlled stability, ensuring safety as their adoption continues to grow.

Regional Approaches to Robot Governance

Different parts of the world are taking their own approaches to humanoid robot safety. China has emerged as a surprising leader in this area. In July 2024, Shanghai published the world's first comprehensive Guidelines for the Governance of Humanoid Robots. These guidelines establish three fundamental principles:

  • Humanoid robots must never threaten human safety
  • They must protect human dignity
  • They must include comprehensive risk assessment and emergency response systems

The guidelines also require manufacturers to train users on the ethical and legal use of their robots – recognizing that safety isn't just about the technology, but also about how people interact with it.

Meanwhile, the European Union's AI Act, which took effect in February 2025, represents the world's most comprehensive AI regulation. With penalties reaching up to €35 million or 7% of a company's annual revenue, this regulation puts serious teeth behind AI safety requirements. Autonomous robots potentially fall into high-risk categories, requiring extensive compliance measures before they can be deployed.

The comprehensive approach of the EU AI Act is examined in EU AI Act: The World's First AI Regulation, which breaks down how this groundbreaking legislation ensures AI systems are safe, transparent, and accountable while setting global standards for AI interaction and regulation.

The Ethics Challenge: Programming Right and Wrong

Beyond physical safety, humanoid robots raise profound ethical questions. Should a robot prioritize saving one person over many? How do we ensure robots don't discriminate against certain groups of people? What happens when a robot makes a mistake that hurts someone – who's responsible?

These complex ethical dilemmas are explored thoroughly in Ethical Dilemma in Robotics, which examines the moral issues that arise with the development and implementation of robotics, from autonomous decision-making to the responsibility and safety of humans versus machines.

Organizations like the IEEE (Institute of Electrical and Electronics Engineers) are developing comprehensive ethical frameworks to address these questions. Their P7000 series of standards projects specifically tackle ethical considerations in autonomous systems. These include standards for creating robots that align with internationally accepted ethical principles and for ensuring that AI systems can explain their decision-making processes.

The core ethical principles that most experts agree on include:

  • Transparency: Robots should be able to explain their actions
  • Fairness: They shouldn't discriminate or show bias
  • Safety: They must avoid harm to humans
  • Accountability: Someone must be responsible for robot actions
  • Privacy: They must protect personal information

Leading companies are also developing their own internal ethics frameworks. Microsoft's Responsible AI Standard, for example, provides a comprehensive 27-page guide that details how to assess AI risks, measure ethical compliance, and manage potential problems throughout a robot's lifecycle.

Microsoft's approach to responsible AI is detailed in Responsible AI (Microsoft's AI Principles), which explains Microsoft's commitment to developing AI systems that are transparent, reliable, and worthy of trust, with principles designed to empower everyone regardless of their backgrounds.

Recent innovations in robot ethics include the development of "robot constitutions" – rule books written in plain language to guide robot behavior, drawing inspiration from Isaac Asimov's laws of robotics. These constitutions leverage large language models to ensure robots understand and adhere to safety principles in complex, real-world scenarios.

This cutting-edge approach to robot safety is explained in Robot Constitutions: Building Safety into AI with Asimov's Laws, which explores how to build safety into AI-powered robots through the creation of constitutions that guide robot behavior using the power of large language models and vision language models.

Real-World Safety in Action

Agility Robotics: Pioneering Safe Humanoid Deployment

Agility Robotics has achieved something remarkable – they've deployed Digit, the only humanoid robot currently working in commercial operations while meeting safety requirements for OSHA-regulated environments. This isn't just a technical achievement; it's a proof of concept that humanoid robots can work safely alongside humans in real workplaces.

The company's comprehensive approach to safety is detailed in Digit at Scale: How Agility Robotics Is Making Humanoids Work, which provides concrete examples of how safety theory translates to practice, including their roadmap for the next wave of ISO safety standards and the complex path to humanoid deployment.

The company's approach involves a three-stage progression:

  • Isolated work: The robot works alone in separate areas
  • Cooperative safety: Humans and robots share space but work on different tasks
  • Collaborative safety: Humans and robots work together on the same tasks

Each stage requires increasingly sophisticated safety systems. The robot needs sensors that can detect humans and animals (yes, even pets!), advanced algorithms that can predict and prevent dangerous situations, and control systems that can react instantly to unexpected events.

The latest version of the Digit robot incorporates several cutting-edge safety enhancements. Among these is an emergency stop button designed to halt all robot movements instantly in case of a malfunction, thereby reducing the likelihood of accidents and enhancing overall safety for both humans and other machines. This feature includes a Category 1 stop, ensuring that power to the robot's actuators is maintained until it can come to a controlled and stable stop, preventing it from toppling over.

Additionally, Agility Robotics has integrated the FailSafe over EtherCAT (FSoE) protocol into Digit. This safety protocol guarantees secure communication over an EtherCAT network in industrial automation settings, allowing for the transmission of safety-critical information, such as emergency stop signals and sensor feedback, while maintaining real-time performance.

Figure AI: Setting New Safety Standards

Figure AI announced in January 2025 that it is building a Center for the Advancement of Humanoid Safety, an in-house division wholly focused on the topic. Led by Rob Gruendel, a former Amazon Robotics safety engineer, this center represents the first industry-wide effort to systematically address humanoid robot safety through rigorous testing.

Figure AI's detailed plan for improving humanoid robot safety is covered in Figure AI details plan to improve humanoid robot safety, which breaks down the company's comprehensive approach to workplace safety and their commitment to transparency in safety testing procedures.

The center will test robots for:

  • Stability while standing and moving
  • Ability to detect humans and pets
  • Safe AI behaviors and responses
  • Navigation systems that prevent injuries
  • Force control to prevent harm during interactions

What makes Figure AI's approach unique is their commitment to transparency. They plan to publish quarterly reports detailing their testing procedures, what worked, what failed, and how they fixed problems. This open approach could accelerate safety improvements across the entire industry.

Practical Safety Implementation

Physical Safeguards: The First Line of Defense

Even with advanced AI and sensors, humanoid robots still need physical safety systems. These include:

  • Traditional Barriers: Guardrails, fencing, and designated robot zones that keep humans safe during certain operations.
  • Smart Safety Systems: Light curtains that create invisible barriers, pressure-sensitive mats that detect when someone approaches, and emergency stop buttons that immediately halt robot operations.
  • Lockout/Tagout Procedures: Critical safety protocols that ensure robots are properly shut down during maintenance. Statistics show that most robot-related injuries happen during maintenance and repair work, not during normal operations.

These essential safety protocols are explained in detail in What Safety Protocols Are Taught In Industrial Robot Training?, which covers the foundational safety awareness, risk assessment training, and hands-on training that ensures workers are confident in their skills when working with industrial robots.

Advanced safety systems like ABB's SafeMove2 allow for collaborative interactions even with large 800kg robots by using scanner systems to adjust a robot's speed based on the proximity of human workers. As a human approaches the robot, scanners send data to the robot control, which then slows the robot down to preprogrammed speeds, coming to a complete stop when a human reaches a certain threshold.

The Human Factor: Training and Communication

Technology alone can't ensure safety – people need to understand how to work safely with robots. This includes:

  • Comprehensive training programs for all employees who will work near robots
  • Clear communication protocols for human-robot interaction
  • Regular safety briefings to address new risks and procedures
  • Feedback systems that allow workers to report safety concerns

Research shows that human error accounts for most workplace accidents, so addressing distractions, workload pressures, and communication breakdowns is just as important as improving robot technology.

Recent Safety Incidents and Lessons Learned

The robotics industry has faced several high-profile safety incidents that highlight the importance of robust safety measures. Recent viral footage of humanoid robots malfunctioning has sparked widespread concern about the safety of robotics technology.

In one notable incident, a Unitree H1 humanoid robot experienced a dramatic malfunction during testing, with footage showing the machine thrashing wildly while suspended from a crane, ultimately causing equipment damage and forcing the crane to collapse. According to the robot handler, "The technical reason is we ran a full body policy while the feet weren't touching the ground. Don't do that". This incident demonstrates how robot control systems can become confused when operating outside their intended parameters.

Another incident involving a Unitree robot during a public event in China showed the robot making an unexpected move toward a crowd of spectators, requiring security staff to quickly intervene. While later explained as a technical malfunction rather than aggressive behavior, these incidents highlight the ongoing challenges in humanoid robotics development as engineers work to refine control systems.

These real-world malfunctions underscore why organizations like OSHA acknowledge that there are currently no "specific OSHA standards for the robotics industry", making the development of comprehensive safety standards even more critical.

Looking Ahead: The Future of Humanoid Robot Safety

As we stand on the brink of a humanoid robot revolution, the work being done today to establish safety standards and best practices will determine whether this technology becomes a blessing or a source of new risks. The collaborative efforts of international standards organizations, industry leaders, and regulatory bodies worldwide are creating the framework necessary to ensure these powerful technologies serve humanity safely and ethically.

The challenges are significant. Companies like Figure AI are aiming to produce 100,000 humanoid robots by 2029, which means we need robust safety systems in place quickly. But the progress is encouraging. From Agility Robotics' successful deployment of Digit in real workplaces to Figure AI's commitment to transparent safety testing, the industry is taking these challenges seriously.

The promising future of robotics technology is showcased in SHOCKING AI That Stunned the World This Month: AI Robot Fights, Nurse Bot, Codex Agent, Veo 3..., which explores how AI-powered robots are transforming industries from manufacturing and healthcare to agriculture and space exploration, demonstrating their potential to improve efficiency, safety, and quality of life when proper safety measures are in place.

Recent developments show robots being designed specifically to help vulnerable populations. MIT's development of the Elderly Bodily Assistance Robot (E-BAR) demonstrates how humanoid technology can be safely implemented to support seniors, preventing falls and providing physical assistance.

This inspiring application of robotics is highlighted in MIT team develops robot that could help older people stay safe at home, which shows how the E-BAR robot acts as robotic handlebars that can support people's full weight, help them transition from sitting to standing, and even catch them by rapidly inflating airbags if they begin to fall.

The future where humanoid robots work alongside us in offices, help care for the elderly, and assist with dangerous tasks is rapidly approaching. Thanks to the dedicated work of safety engineers, ethicists, and regulators around the world, that future is looking increasingly safe and promising.

The broader benefits of robotics technology are explored in Robots for the SDGs: Leveraging AI-powered robotics for good, which demonstrates how autonomous robots are advancing the United Nations' Sustainable Development Goals through applications in agriculture, disaster management, healthcare, education, and humanitarian aid deployment.

As these remarkable machines transition from science fiction to everyday reality, one thing is clear: the rules and standards being developed today will shape how humans and humanoid robots coexist for generations to come. The stakes couldn't be higher, but neither could the potential rewards for getting it right.

The development of humanoid robot safety standards represents one of the most important engineering challenges of our time. As these technologies continue to evolve, ongoing collaboration between industry, regulators, and the public will be essential to ensure they enhance rather than endanger human welfare. With proper safety measures in place, humanoid robots promise to revolutionize how we work, live, and care for one another, creating a future where humans and machines collaborate safely and productively.

Media Gallery