Perhaps you’ve seen videos of Theo Jansen’s Strandbeest, or “wind walker,” and have decided that you need your own. This happened to me several years ago, and after four ‘beest iterations, I finally have one that works fairly consistently (and was featured in Make: volume 54!). It made it through most of two days of walking around at the Orlando Maker Fair this year, and seemed to attract a lot of attention, so I would consider it extremely successful.
That being said, it took several years from inspiration to having a working model in my garage. Jansen has published the linkage lengths needed to duplicate his machines, but as I have found out, that’s just the beginning. I learned quite a bit while making these, so here are a few things that could help you if you decide to take the journey to make one.
It’s a Big Job
First of all, you’re likely not going to design and build one of these contraptions in a weekend. I started out trying a ‘beest with four legs, and found that this was much too unstable this way, however, when powered, it did make a nice display while suspended. Even getting to that point took nearly a year of on-and-off work. The next one, which had 8 legs and was about the size of a golf cart when assembled, never really worked. Apparently I had bit off much more than I could chew.
The next two iterations (the first of which is seen here) were much smaller and lighter and actually walked. Apparently my “prototype” was actually quite a bit larger than the finished product.
Precision and “Slop”
Though Jansen’s machines were mostly built from PVC pipe, if you want to manufacture your own, everything still needs to be fairly secure. Where this is especially evident is in this type of walker’s leg assemblies. If they can flop around at the joints, a few degrees one way or another really adds up by the time the legs touch the ground. The central camshaft assembly is another problem area, as it can get bound up if not properly aligned.
To combat this in the legs, they need to be wide enough to not “flop” significantly, or be held together relatively tightly. Either solution, unfortunately, can have consequences in increased weight or friction, so you must consider the tradeoffs of each. One idea I’ve seen is using spacers inside of the leg joints to widen them without adding as much weight, which seems like a good solution.
Friction is a Killer
Unlike a robot design that has, say, two motors attached directly to two wheels, if you’re transmitting force in a Jansen mechanism there are many, many areas for friction to multiply. Depending on how you calculate it, each leg has seven individual joints that are both rubbing against some type of shaft, and against each other. Add in a huge amount of friction in the traditional design’s cam shaft, and you have some serious losses.
In one of my designs, I tried to compensate by using carbon as a dry lubricant, but in my fourth — successful — ‘beest I actually used thrust bearings in many of the joints. At perhaps a couple dollars each, the price of them can add up very fast, but some means of reducing friction is generally needed while keeping proper alignment. Teflon washers might also be an acceptable substitute.
Torque Adds Up
Although Jansen’s original ‘beests feature a sort of PVC camshaft design, by the time you get out to the second, third, or even further out set of legs, the torque applied to this linkage can really add up. I used this design in my third ‘beest, but by my fourth I decided to do things a little bit differently. Instead of using a cam shaft, I used a central shaft with gears on it, similar to what’s see here in the video above by Joachim Punkt.
For me, this worked very well, but I still had to file the gears down quite a bit, as I didn’t design enough “slop” in them to work well. Interestingly, in machine design, it seems that loose fits can be very good in some cases, but works poorly in others.
Computerized Tools Help
For my first ‘beest, I cut everything by hand. Though it worked, it limited my build options, and took a lot of time. The next three iterations were cut by a computer. This meant better accuracy, and many more options as far as the design went. Also, cutting everything by hand, if I could cut it accurately, would have taken a very long time.
Assembly takes long enough, and your time is worth something. Depending on your situation, don’t be afraid to get help from others, even if that includes a computer!
Select Appropriate Motors
As larger motors tend to be heavy and quite expensive, the motors I first used, pictured above, were quite undersized. Though they were much too small for the application, after breaking a few, I finally was able to make the legs rotate while hanging. I also tried windshield wiper motors on ‘beest 1 and 2, but wasn’t able to control the rotation without losing a significant amount of torque. The lesson being that if you want a large motorized walker, plan on paying hundreds of dollars or even more for a very good set of motors.
When I built my third iteration of the ‘beest, shown above, I used very small motors, powered, in fact, by a quadcopter controller. The fourth used the same type of motors that I initially tried on the much larger “beest 1.” In this application they worked quite well!
I haven’t yet been willing to pay the price, both in time, materials (including motors, bearings, structural materials, an storage space) to build a proper “golf cart walker,” but others have been successful with larger models, Jansen being the first. On the other hand, if this seems like too much work, perhaps one of these in kit form could be a good compromise if you’d like to have your own.