Safety Instructions and Protocols for AMR and Collaborative Robot Integration

    Hi everyone,

    I am a member of a research group focused on the integration of Automated Mobile Robots (AMR) and collaborative robots. Currently, I am concentrating on safety instructions and Protocols at various levels, from shoulder-to-shoulder human and machine operations, such as deploying 2D lidar, to high-level programming (ROS integration) of six-axis robots, including DCS safety measures.

    Imagine an AMR in a sterilized environment assisting operators with specific tasks. How should the collaboration safety between humans and AMRs be managed? What safety instructions should be in place between AMRs and other robots or machines? What considerations are necessary?

    I plan to conduct an extensive literature survey on this topic. However, I am currently encountering a lack of related sources(book, paper, link, keywords, ... ), as most of these safety instructions and detailed approaches are proprietary information specific to individual companies and projects.

    Therefore, I am reaching out to the community for guidance on potential resources related to this topic. Any hints or references would be greatly appreciated.

    Grazie

    Edited once, last by Rick petrous (June 3, 2024 at 11:02 AM).

    Well, I'd start with the actual safety standards: https://webstore.ansi.org/standards/ria/…tiverobotsafety
    ANSI/RIA R15.08-1-2020 - Industrial Mobile Robots - Safety Requirements - Part 1: Requirements for the Industrial Mobile Robot

    These are the standards installs are judged against by the relevant legal authorities, or compared to in the event of an injury lawsuit.

    Of course, there's a lot of variables. A collaborative robot with a chainsaw attached is no longer collaborative. But a cable attachment bracket mounted to the side of a cobot's arm can also potentially make the robot non-collaborative, if it has any sort of sharp-ish corners on it -- there's a reason that all cobots are swoopy curves that look like something designed by Apple.

    The RIA specs center around three questions:
    1. How easy is it for a hazard to occur?
    2. How much harm can the hazard cause?
    3. How easy is it to avoid the hazard?

    The combination of the three answers creates a risk factor, which then must be mitigated. Some facilities have risk mitigation strategies that amount to "tell people not to be stupid" and don't bother with anything else. Others have internal safety standards that go well above and beyond the RIA minima.

    For cobots, the hard minimum is impact force over contact area -- the same force may be tolerable for a "blunt" contact but intolerable for a "sharp" contact.

    For AMRs, it's all about sensing, and being able to stop safely before collision (and also without, say, letting a heavy load slide off the AMR and fall on someone during an E-Stop).

    Hi Skyfire,

    Thanks for your answer and the example of apple-rounded robots. That's exactly the kind of practical experiences I'm looking for, as they are often unwritten or hard to find. Unfortunately, I don't have access to ANSI/ITSDF standards due to a lack of funding for the literature survey phase. Is there any way to access that material?

    I also wanted to highlight your last point about machine-human communication. The versatility of automated processes means that safety structures can be interpreted differently, so I'm looking for a general description (apart from ANSI/ITSDF) in research or practical resources. This aspect remains unresolved for me.

    Thanks again!

    Quote from Rick petrous

    Unfortunately, I don't have access to ANSI/ITSDF standards due to a lack of funding for the literature survey phase. Is there any way to access that material?

    Mmm... good question. I'm sure there are illegal copies floating around the dark web, but that's obviously not a great thing to put in your bibliography.

    As the standards get updated regularly, you might be able to find a legit copy of a slightly outdated version. Which wouldn't be ideal, but the year-by-year differences are usually pretty minor. And the project lead times are enough that most hardware being built "today" is based on the standard from 2-5 years ago.

    I've never pursued this option, but I have to wonder if there are any public or university libraries that keep these standards on the shelf.

    There are large companies that use lots of robots that publish, at least for their vendors, their own in-house versions of the RIA standard, which generally consists of the relevant subset of the RIA standards, "dumbed down" for easier application. GM, for example, has an enormous book for this, but where the RIA standard says "Robot stopping distance for size/payload/speed must take the fence into account", GM just says "for robots in this weight class, allow X inches of stopping distance universally and forget the details" (Hence my "dumbed down" comment), because GM has millions of runtime hours of data to back that up, and their house standard exceeds the RIA standard in most respects. You might be able to find such companies willing to share their house standards to a research team -- not as good as getting the actual RIA spec, but if you could collate multiple house standards for their common elements, that could still be useful data.

    Quote from Rick petrous

    I also wanted to highlight your last point about machine-human communication. The versatility of automated processes means that safety structures can be interpreted differently, so I'm looking for a general description (apart from ANSI/ITSDF) in research or practical resources. This aspect remains unresolved for me.

    It's not an area I've delved into in any detail, but all Cobots I've seen have some sort of light (usually integrated into the arm somehow) that indicates, at minimum, whether the drives are energized or not. Multi-color LEDs with different functions seem to be the current rage, with every color/flash/breathing/whatever pattern indicating something different. Likewise, most AMRs I've dealt with have some sort of colored lighting to indicate moving/stopped/waiting/etc. How much of that is driven by the RIA AMR standard vs just UX/UI factors I'm not sure.

Advertising from our partners