Living Robots: Revisiting BEAM

You’re hit by the global IC shortage, reduced to using stone knives and bearskins, but you still want to make something neat? It’s time to revisit BEAM robots.

Biology, electronics, aesthetics, and mechanics — Mark Tilden came up with the idea of minimalist electronic creatures that, through inter-coupled weak control systems and clever mechanical setups, could mimic living bugs. And that’s not so crazy if you think about how many nerves something like a cockroach or an earthworm have. Yet their collection of sensors, motors, and skeletons makes for some pretty interesting behavior.

My favorite BEAM bots have always been the solar-powered ones. They move slowly or infrequently, but also inexorably, under solar power. In that way, they’re the most “alive”. Part of the design trick is to make sure they stay near their food (the sun) and don’t get stuck. One of my favorite styles is the “photovore” or “photopopper”, because they provide amazing bang for the buck.

Back in the heyday of BEAM, maybe 15 years ago, solar cells were inefficient and expensive, circuits for using their small current were leaky, and small motors were tricky to come by. Nowadays, that’s all changed. Power harvesting circuits leak only nano-amps, and low-voltage MOSFETs can switch almost losslessly. Is it time to revisit the BEAM principles? I’d wager you’d put the old guard to shame, and you won’t even need any of those newfangled microcontroller thingies, which are out of stock anyway.

If you make something, show us!

39 thoughts on “Living Robots: Revisiting BEAM

  1. People had technology on this continent before Europeans came over. Canoes and kayaks, nets for fishing, baskets made through complicated and time consuming methods, I saw one recently that is 200 years old. There were platforms over the Columbia River for fishing. In the north, they built igloos, an integrated life in a harsh environment, while modern housing has problems.

    When people came over from Europe, they brought other technology, which could easily be adapted because those metal knife blades were not so different from the knives already being used. And trading was a familiar concept.

    People like Spokan Garry could go from the northwest in 1825, where the only Europeans were fur traders, to school in Red River, to learn English and other things, and do it successfully.

    If you can’t get your 555s, there is a whole lot of technology, endless layers, before you get to “bearskins and flint knives”

        1. The annoying part is, some people will call you ignorant, then others will call you old when you know stuff they don’t (or looked it up, they say you must be xxx). You just can’t win.

  2. I loved all the “misuse of electronics” that went into BEAM.
    Some of those circuits are just crazy. Lovely, but crazy.

    I see Solarbotics still sells a bunch of Solar Engine stuff for cheap.
    https://solarbotics.com/?s=solar+engine&post_type=product&dgwt_wcas=1

    I’d like to see new designs also. It might be a great summer camp project for kids.
    Post links if you find any please. Or contact me at firstlast at gmail if you want to collaborate on curriculum.

  3. Mark had an office on the 3rd floor of the Math and Computer building at UW when i was an undergrad.

    I will never forget Mark showing me his photovores, photophobes and solar rollers… reusing an annoying birthday card as a controller.. his amazing skill with wire and solder…

    Today we take for granted that a quick search will return lots of “Mark Tildens” but i was lucky back in 1990 to cross paths with Mark and it had a stong influence on me.

  4. First off, thank you for trying to re-ignite this deep and quirky bit of robotics history instead of just surface-level reposting a solar-engine or LED pummer. There’s much more to be found in the form of bi-directional motor driving, all-analog neural networks, and stateless multi-legged robot motion control all working from the most meager of power sources quite happily.

    There are some quality quotes from the progenitor of B.E.A.M. Mark Tilden scattered across dusty sites and publications that are getting harder and harder to find. This article pushed me to dig through a few and Mark’s reasoning for why B.E.A.M. is worthwhile seems pretty relevant in these lean times:

    >The secret to BEAM technology is performance to-silicon ratio. If you had to basically restart the entire computer robotics industry on a desert island, this would be the technology you would use. This is the secret that [John] Von Neumann, and [Michael] Brady, and [Alan] Turing, and everyone else was looking for. Asynchronous real-world control in the smallest number of possible transistors to the maximum possible effect.

    B.E.A.M. was such a inspiration when I was a kid. The concept of these little devices built from spare bits and pieces harvested from junked electronics, given the ability to move, find energy, and deal with obstacles in incredibly robust ways was just so neat. Given that most of the field has shifted to a top-down all-digital designs, they still are.

    That said this isn’t a panacea for brittle robotic design but it was a nice meeting of requirements and solutions somewhere in an achievable middle; don’t try to make a robot that can do every complex thing imaginable from a central CPU, stick to simple needs like self-perseveration and build from there, spreading out the “thinking” across the entire design in the hardware itself. Not to mention the idea of using electronic trash was icing on the cake for a kid who enjoyed taking busted electronics apart.

    I’ve had a glossy photo book at my desk for years called Robo Sapiens that contains a short interview with Mark Tilden about his B.E.A.M. movement and his thoughts on robotics circa 1999-2000. His opinions regarding robotics in the home, the inflexibility of digital systems in complex environments, and the fragility of common robotics still ring true. Especially when he digs out an 11 year old BEAM design for a NV-net controlled snake robot and, despite having multiple mechanical failures, it still can effectively locomote. The photographer was so enamored with Tilden’s walking designs that both the first and last images in the book are of his Unibug 1.0 puttering happily along the uneven terrain at the Sand Dunes National Monument in Colorado; images I still find energizing.

    Sadly, Solarbotics has let link-rot and site-spread ravage their 90s-era repository of ASCII formatted circuit diagrams along with letting the tiering structure of what circuit came when fall into disarray at the hands of slipshod geocities-tier subsites.

    I’d love to see the solar engines, NV-cores, and motor drivers get a modern overhaul. Getting a lot of the key pieces for many B.E.A.M. circuits is getting nearly impossible and I’m pretty sure that they’re not best-in-class for what the circuits are trying to achieve anymore.

    ——–

    P.S. If anyone has information on Mark Tilden’s solar powered B.E.A.M. window scrubbers that he had running for many years at his house before moving to Hong Kong, I’d really love to learn more about them!

    1. Yeah seriously I need a HaD article on those NV-cores.
      A robot lizard with only two motors and a PLL (or whatever it was) that can adapt to introduced terrain? Lost art for sure.

      1. The Nv cores challenge… My own experience with them is kinda mixed.

        For those following at home, the idea is that you have an oscillator, and you have various ways of interconnecting them that change/couple their duty cycles, and that leads to changes in behavior.

        So like, the left/right motors fire on a bot on the high/low parts of the cycle, and the relative duty cycle is influenced by light sensors. This gives you light seeking or avoiding, depending.on which way you wire it up.

        More complicated behaviors could come out of layering these influenced-oscillator systems on top of one another and having them drive each other. It’s _not_ unlike how simple invertebrates work, so it’s not crazy. But it does leave you with the task of designing an invertebrate nervous system from scratch, and then calibrating it. It’s worth noting that it took Mother Nature millions of years to do this.

        The only one of these bots I ever got working were those simple types — photovores / photopoppers / FREDs. I must have built three or four, though. My crab-bot is probably still in the closet… But when it comes to the more complicated hierarchical systems, they always required a lot more tweaking and, I suspect, deep analog knowledge than I had at the time.

    2. “P.S. If anyone has information on Mark Tilden’s solar powered B.E.A.M. window scrubbers …”

      Million-dollar idea: solar powered gutter de-leafer bots. They can move essentially infinitely slowly, can be ugly b/c they don’t have to be seen. Design challenges are weather-proofing, traction, and of course cost.

      1. Multi-billion dollar idea: a robot to scrape barnacles from the hulls of seawater ships. This is the single largest line item in the US Navy budget. Whoever makes a practical solution will quickly become the world’s richest person.

  5. Yes, I loved this approach to electronics and the creativity in the BEAM idea. I had a few twitchy bots hanging in the windowsill and a ratty folder with diagrams and ideas for new bots. Remember making those free-form “H-bridge” controllers. The Junkbot book was great. I would LOVE if a site or a book with modern bits and parts was created. Fun way to learn to solder as well (using paperclips and thumbtacks.

      1. Yeah I’m having a hard time finding a video of them but the computer whiz that designed the combat robot had a bunch of them crawling around in his trailer.
        And Mark Tilden was brought in as a technical consultant for the film. Though I can’t find the original website that I read on that either.

  6. BEAM robots will always have a special place in my heart, I spent so much time looking them up in the school computer lab in the early 2000s. A number of the ones I built still work today. I wish there were more info on the advanced walker robots Tilden made, but it seems like much of it is lost.

  7. wow, those pages are still around? that great stuff, i still have my solarbotics kits, some unopened. the old glass solar cells, 1381’s and many bits and bobs

  8. I have many happy memories of seeing BEAM & Solarbotics in action at Robotix 96 or 97 in Glasgow. It kickstarted a great little maker hobby of mine. Glad to see that it’s still around.

  9. Solarbotics – we’re still around, and even still dabble in new BEAM kits. We have a new prototype going through burn-testing right now and should have it ready for this fall. As a company, we’ve branched into more traditional streams, and now are pretty much a “Mom & Pop Electronics Shop” more akin to what Radio Shack used to be. I just had this discussion recently where the topic was the relevance of stores like ours. My position is the modern barrier to entering electronics is /so/ much easier to thanks to Arduino and pre-engineered modules, I wager that it’s now so much easier to build your own solution with modules & example code there will remain to be a need for custom projects. Curating what we should carry in the store has been fun.

    Regarding BEAM popularity, there was that moment a more-than-a-few-years back when the threshold of BEAM cost-effectiveness was breached. The cost of setting up a 74HC240 with capacitors and trimpots was somewhat more than a single ATTiny. Not having that cost advantage over $80 68HC11 & $400 compilers anymore did take the wind out of the BEAM sails somewhat. Still, setting up phase-relationships between different oscillator circuits is _so_ much more tangible when you’re twiddling a trimpot rather than plugging in values. Watching effective walking gaits come out of such simple systems gave me huge respect for the evolutionary process.

    Solarbotics.net may be going unmaintained, but it hasn’t gone unforgotten. We’ve spent some effort in making sure that the archive is still somewhat active as we’ve moved our own site through various providers. Sorry if there’s crust on the pages, but at least it’s better than being totally gone!

    1. AMEN.
      May BEAM live on!! The idea of making something that does not require a program, yet reacts to its environment and appears to move intuitively is why BEAM is so cool. A bit old school? What else can be done that has not been explored? Have we sold the idea a bit short and given up?

      Peter B T

  10. I am also a big BEAM fan. Mark T, solarbotics, and so many others. I used the BEAM kits to teach so many people the basics of biology principles, electronics, design (Aesthetics) and mechanics.

    There some things the BEAM nvs Mark built I like more than Deep learning a lot of the low level.

    A new thing I have been playing with is for a reasonable price and doing Machine learning and deep learning at a affordable price and portable device..
    https://luxonis.com/

    I always think of merging the two BEAM and compute base learning.

    All the best to you all! Keep learning and sharing!

    Mark D

  11. I LOVE the idea of BEAM robots. Wondering though whether more could be done on a larger scale. I am interested in researching whether neuro networks made using RC networks and NOT gates etc can still find a place in the modern world to do real useful tasks.

    Peter B T

Leave a Reply to Elliot WilliamsCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.