While some humanoid robots we've reported on recently can walk up stairs or fight fires, researchers at the Tokyo Institute of Technology have developed a "swumanoid" robot they say faithfully mimics human swimming motions for applications that may include more streamlined swimsuits or methods for improving the performance of competitive swimmers. Other uses that come to mind include robots as an aid in physical therapy.
The half-scale swumanoid incorporates 20 waterproof motors that are controlled by a computer to mimic the actions of human swimmers. Tokyo Tech researchers Chung Changhyun and Motomu Nakashima say they developed the robot by carefully analyzing real human swimmers' body movements.
Previous research relied on analyzing video images of human swimmers to monitor their movements, but the researchers say this led to the problem of how to reproduce those movements. That's because this approach depends on the ability of the swimmers to recreate their own movements precisely. It also depends on the ability of researchers to make accurate measurements of the forces that are acting on swimmers' bodies while they are in motion. Both of these tasks have proven extremely difficult to accomplish.
The half-scale swumanoid swimming robot incorporates 20 waterproof motors that are controlled by a computer to mimic the actions of human swimmers. (Source: Tokyo Institute of Technology)
Nakashima and his colleagues at Tokyo Tech previously developed the SWUM (SWimming hUman Model) swimming human simulation model for analyzing the whole-body dynamics of swimmers. Data for body geometry, joint motion, and specific analysis settings are given. The model estimates the fluid force acting on the swimmer's body, solves the equations of motion of the rigid human body, and computes the absolute motion of the whole human body. The developers have created a free implementation of the model called "Swumsuit" that can be downloaded here.
Two of the major issues to resolve have been reducing body resistance to movement through water and increasing the driving force caused by the swimmers' hands and feet. The swumanoid robot enables control of the precise movements and driving forces of the robot swimmer. The robot is half the size of an adult human, but has the same proportions and external features. Twelve of the 20 motors actuate the movements of the arms and one motor turns the waist. Each arm has six degrees of freedom, while the shoulder has four degrees and the elbow has two. Extra joints have been added to the shoulder to reproduce the retraction of the shoulder blade.
The swumanoid demonstrated the accuracy of both hardware and software models by its performance in reproducing the arm motions of the front crawl, which is the most complicated swimming stroke.
Changhyun and Nakashima say that in the future, they want to produce much faster swumanoids and also analyze those robots' movements to develop high-speed swimwear.
The robot definitely mimics human form, but it still appears choppy in its movements--no where near as fluid as a human swimmer and not even remotely comparable to a competitive one. That said, I get the idea of applying the robot to aid in physical therapy exploration, but don't necessarily see how it would aid in helping competitive swimmers.
All of that said, between this and your earlier post this week on marine-inspired robots, there's definitely quite a lot of innovation and progress going on in this field. Thanks for sharing.
I agree, the robot is not yet perfectly human in all its movements. But it's awfully close, and compared to previous efforts, this one's elegant and fluid. The main achievement isn't just the robot that demonstrates the SWUM model, but that model itself, to emulate all of the human motor movements accurately, get the "map" done, so they can then be refined even more. But first you've got to have the entire, accurate map. Once you have that accurate map, you can also use it plus the analysis to map human swimmers' individual movements and analyze them, presumably leading to greater efficiency.
Rescue robots are currently being developed mostly for use on land. Based on some other nautical robot designs, I'd guess that rescue robots developed for use in the water would not replicate the details of human anatomy or swimming movements, as the swumanoid does, since human swimming movements aren't the most efficient way to move through water. The swumanoid has been optimized for those last two functions, but rescue robots are optimized for speed and strength, such as lifting or pulling heavy objects. For example, the Hawkes Remotes T-Series is small, compact and provides enough torque to lift 220 lbs: http://www.designnews.com/author.asp?section_id=1386&itc=dn_analysis_element&doc_id=246206&image_number=11
Having watched hundreds of lifeguard rescues using jet skis here on the Southern California beaches, I don't think any humanoid shape could match that performance. Watch how jet skis take surfers out to huge waves during professional competition, then speed out of the way of the huge wave before it breaks. Also, a skilled person paddling on a surfboard is faster than any human swimmer, as lifeguards also use surfboards for doing rescues.
This research is aimed 100% at sports-related apps, as stated in the article (competitive swimmers, optimized swimsuit design for same). Not at anything as practical as search-and-rescue, or replacing lifeguards.
Beth. I agreee with you. When I first saw the video, the first thing that I noticed was that the movement was choppy. Why they didn't put sensors on a swimmer to provide a model is beyond me.
The other thing that I noticed was that the arm movement was wrong. You have to do two things at the same time which the robot seems to not be able to do. You pull down one arm at the same time you move the second arm up. It appears that they move the one arm up AND THEN move the other arm down and back.
I would bet that if they put it in water that it would sink which is why the video doesn't show the robot actually swimming.
Great story Ann!! That is so interesting and love the video. If this robot has the ability to pick up things this would be great for underwater rescues. It would be nice to see some of the other robot teams work together to combine their specialty features into one robot.
Do you know what's the teams next step for this robot?
Thanks, gsmith120. That's an interesting idea, to combine these swimming abilities with some of the abilities of the robots shown in the nautical robot slideshow. The team stated that its next steps are to develop faster robots to better emulate competitive swimmer's movements and also to come up with swimwear optimized for high-speed swimming.
Using the robot to enable competitive swimwear--that's likely to open up the can worms of too much reliance on technology and not enough on performance of the human body. Nevertheless, as Ann points out, the fact that they put all this energy into developing the full swimmer's body model is exciting and an effort that could have applicability in numerous places.
I think the point of optimized swimwear is to not slow down the competitive swimmer. Any clothing does so: it introduces drag. Regarding arm movements, it's important to remember that this hardware is a first effort, although the simulation software is not, and the researchers point out that getting the shoulder movement down was the hardest part, as it's the most complex. It's a WIP.
As someone who has spent more swimming hrs in the water than 1000 average people combined I found this a poor use of everything. No wonder Japan is falling far behind.
Next the human form and especially this unit is about the worst you can get for propulsion movement in the water as can exert little force with a very large amount of drag.
I hunt fish. etc for food underwater, no scuba, just snorkeling to 50' down and even with fins I'm almost the slowest fish down there except the poison ones. Only by tactics is one able to get the fish one wants to eat like sneaking around a coral head, etc and surprising them.
My thoughts as well. Why look to the human form for a model of a swimming platform? Fascinating project and quite a feat to be able to accomplish, but why not look to a more efficient model for swimming than the human form?
Brentlim, as we mention in the article, this research is not theoretical, designed to come up with a generic swimming platform. Instead, it's 100% targeted at sports-related apps, specifically competitive swimmers, and optimized swimsuit design for same.
This is a great start. Agree with Beth that its movements are quite choppy. I also noticed that it has the advantage (or disadvantage for good human swimming simulation) of not having to raise its head above water to take a breath. We all could swim a bit better if breathing was unnecessary. Seems like simulating taking a breath and seeing where you are going must be a part of this at some point.
I think it would be a good idea under certain conditions like a bad weather where it's almost impossible to send someone to rescue, but I think the legs in water are inefficient and for that a propeller for lower body is much better in many aspects. The upper body with limbs is a good idea if you are trying to grab someone or something from the water. Nice work though!
This is a fascinating article and a good report but I must agree with Jerry, it seems to be a misuse of technical resources; i.e. time, money, CFD, etc. Then again, sometimes the greatest break-throughs comes from seemingly trivial pursuits. I think this exercise must be aimed at other than saving lives. I would like to know more about the control methodology and what language was used to program the device.
Yes, this is basic research stretched to its limits. I guess we never know when something like this could serve as a foundation for a breakthrough in another area, but it's hard to imagine what it could be.
Hello Ann. I definitely will. Robotic systems fascinate me and each year ( if not each month ) there seem to be advancements that just amaze. I definitely will take a look at SWUM and see if I can gain additional information; then I'll certainly will let you know.
With a better understanding of materials’ response to load and temperature, researchers could potentially use the knowledge to improve design. The research could even help geologists studying plate tectonics.
Focus on Fundamentals consists of 45-minute on-line classes that cover a host of technologies.
You learn without leaving the comfort of your desk. All classes are taught by subject-matter experts and all are archived.
So if you can't attend live, attend at your convenience.