A 3D rendering of a humanoid robot working on a digital display

Safety Rated Data Communication for Humanoid Robots

There are currently no dedicated functional safety standards for humanoid robots. Therefore, for now at least, you have to rely on the guidance from the available standards, including IEC 61508 (general industrial functional safety), ISO 10218 (industrial robots), ISO 13849 (machinery), IEC 61784-3 (black channel data comms), etc. I think humanoid robots might have to rely more on IEC 61508 going forward, and I will cover that in a future blog. In this blog, I will discuss safety-rated data communications for industrial robots, including humanoids.

The main industrial robot standard, ISO 10218-1:2025, has a new section 5.3.6 covering communications. It is a short section but contains several significant ideas. This clause discusses "transmission category," which is an idea borrowed from IEC 62280/EN 50159, a rail communication standard.

Transmission category 1 is a closed transmission system - Think of the internal network within a robot.

Transmission category 2 is more open - Think of an Ethernet cable between a fixed robot and a PLC

Transmission category 3 is the most open category – it might be a WiFi network connecting a humanoid to a centralized control system

Then, depending on the transmission category, more or less rigor is required against the usual networking threats, as specified in standards such as IEC 62280 (rail communications) and IEC 61784-3 (fieldbus), both of which are referenced in IEC 61508.  The importance of the defenses (countermeasures) against the common data communications threats is emphasized as follows

     "-" - not really important since unlikely for this transmission category

     "+" - important do something to detect these failures which are likely for this transmission category

     "++" - very important, try very hard to protect against these threats for this transmission category

Table 3 from ISO 10218-1:2025 is shown below and shows the defenses against the commonly recognized communication network threats for each TC (transmission category).

 Transmission category from ISO 10218-1:2025

Figure 1: Transmission category from ISO 10218-1:2025

This means that for a network carrying safety data within the robot body (TC 1), a lot of effort should be put into protecting against data corruption (see an earlier blog here). For the other defenses, you need to do something, but a lower effort is sufficient. ISO 10218 treats masquerade as not important (‘-’) for TC1 (i.e., additional measures are typically not expected).

In this context, masquerade means a node on the network that looks like a valid safety‑related communication node, and the receiving system cannot tell the difference unless specific counter‑measures are implemented. However, for an internal robot network with a fixed number of participants (for instance, one per axis, one for the robot controller, and one for an end effector), and with the robot’s body preventing the addition of new nodes, problems due to a masquerade threat are deemed unlikely.

 Description of a masquerade error from IEC 61784

Figure 2: Description of a masquerade error from IEC 61784

For the more open/less well-defined TC 2 more of the threats need to be taken seriously including masquerade threats for TC 3. For TC 3 you have an open network such as a wireless network with a potentially unlimited number of participants.

Note 1 – masquerade does not include an intentional effort to deceive, which would be cybersecurity, and cybersecurity is covered by the new section 5.1.16 in ISO 10218-1:2025.

Note 2 – threats is a term more normally used in cyber security standards, but is used here in the context of what IEC 61784-3 calls “communication errors”.  IEC 62280 does use the threat terminology in a non-cyber security context.

The defenses above are typically implemented in an SCL (safety communication layer) at the endpoints of communication links. The SCL is typically software developed for an SC (systematic capability) and running on hardware developed with an SC, and typically contributes no more than 1% per logical link to the PFH (average probability of dangerous failure per hour) of the safety function.

The “+” and “++” require different levels of effort. A higher level of effort might be characterized by having two defenses against a threat. For instance, if considering “delay” as the threat, for a threat marked as “++”, you could have a timestamp and a timeout, whereas only one defense if the threat was just “+”.  

For repetition threats, higher effort might mean a larger sequence number rather than just toggling a single bit. The higher sequence number offers more protection against a roll-over of the code.

Coming back to humanoids, their mobility means some form of wireless networking. Any safety-rated communications on such networks will therefore require high effort against all of the threats if you choose to follow ISO 10218:2025. This makes it more complex than for a fixed robot, which could be networked over an Ethernet cable. In an ideal world, a humanoid would not rely on the wireless network for safety, but I’m not sure that will be possible.

Assuming you got to the end of this blog, I invite you to check back on the second Tuesday of next month for the next blog in this series. Until then, I hope to post “mini blogs” on the other Tuesdays in the month directly from my LinkedIn account. Please follow me on LinkedIn if interested.

Related Blogs

1. Functional Safety for Networking

2. The Revised Industrial Robot Safety Standard – My Highlights

3. The Most Relevant Standards for Industrial Robot Safety

For previous blogs in this series, see here.

For the full suite of ADI blogs on the EngineerZone platform, see here.

For the full range of ADI products, see here.

Before You Switch


Switching languages will make ADI Explorer unavailable. Resume your session by switching back to English and reopening ADI Explorer.