A YouTube video released by Boston Dynamics is making some waves today. They decided to unleash "Wildcat" for the public to see.
Many are already familiar with the slow and lumbering "BigDog" that has been in development for the last few years. We've seen it take steps to avoid obstacles and wander around with surprisingly lifelike motion. We've even seen it recover from falls and tumbles, slowly lifting itself back up and trudging onward. However, we were always able to relax a little bit, knowing that at least we could outrun this four-legged beast, should the need arise.
When Boston Dynamics showed videos of its "Cheetah" quadruped system that is patterned after the feline that shares its name, we were able to relax a little bit knowing that, even though it could theoretically move faster than any living human being, it was confined to the treadmill due to its system of hydraulic tethers.
Wildcat offers no such respite. As you can see in the video, this bounding 'bot could easily catch all but the fastest of us mere mortals. Information is very sparse right now, since Boston Dynamics hasn't provided its website with any details yet on its latest creation. The only information we have so far is from this video description:
WildCat is a four-legged robot being developed to run fast on all types of terrain. So far WildCat has run at about 16 mph on flat terrain using bounding and galloping gaits. The video shows WildCat's best performance so far. WildCat is being developed by Boston Dynamics with funding from DARPA's M3 program. For more information about WildCat visit our website at www.BostonDynamics.com.
Click this image to see a selection of still shots from the video.
Judging by their past experiences with BigDog and the massive improvements it has made over time, I would expect some uneven terrain demos at some point. However, just like everything else that runs on legs, it will obviously be slower in unpredictable circumstances.
I could be wrong here, but I think the goal is to have something that can transisiton from sprinting on even terrain to navigating things where wheels don't work well, Imagine this and bigdog combined.
The legs are lean relative to its body for the same reason horse, deer, and other quadruped mammals have lean legs. The relatively short proximal hindlimb has a shorter range of motion but is driven by extremely powerful muscle groups and large tendons. The lean distal hindlimb translates the short movement of the proximal hindlimb into a large movement at the hoof, so is mostly bone, requiring far less muscle. The same is mostly true for the forelimb. The light distal limb requires less energy to move, resulting in very efficient locomotion.
Interesting they are using the gasolin engine for a powerplant (hydrolics?) - usually proof of concept doesn't need to deal with a "practical" power source. This thing could probably have a large range using gasoline. It was nice to see the gallop and trot - any dressage ;) We have seen others climb stairs and clear obsticles although who really knows what it's capabilities/limitations 16mph is pretty quick on legs. That is quite a large mass to move, balance and even turn - very, very impressive - and it seems to have all the logic on board - it would be nice to know how autonimous it is. It also looks like they could easily tuck the legs, or have it in a cart for transportation. Very impressive demonstration.
This is great progress. Sure it is running on a flat surface, but you have to crawl before you can walk (or run rather). Military uses aside, this is really a technology foundation for animal and/or human like assistance robots - think of the movie "iRobot". We may even see robotic pets like dogs, cats, perhaps even horses. Maybe someday replace the bulls and horses at rodeos removing the life threatening aspect? Could replace the fox in the English "fox hunt"? Could be used as horse riding training vehicles - much safer for children to learn that way? No more greyhound racing using real dogs - contestants compete instead with their latest greyhound robotic dog? I wonder why they used a combustion engine (I'm guessing some modified motorcycle), instead of an modified electric motorcycle. Maybe that's what they will do next?
You might be right Caleb...but looking at bison, deer or a goat I get the impression that they do have more powerful legs although I didn't make any measurements ;-)...artificial legs I guess are made from stronger material so that is why they probably be so thin...still wildcat looks obese and not esthetically pleasing! (not that it matters to the application)
Insect legs? I wouldn't worry too much about size, when comparing the power of mechanical legs vs the legs of a mammal, for example. And as to worrying about the speed? Man has created any number of "creatures" which can move faster than their creator. Not to mention that there are any number of four-legged creatures which can do so anyway.
Great demo. Seems to me that expecting this device to move as fast in rough terrain as it does over flat terrain is not essential. After all, the same laws of physics apply to it as to living creatures. You wouldn't expect a cheetah to run as fast over rought terrain either. Rough terrain creates faster accelerating movements in different directions, at any given horizontal speed, compared with flat terrain, so you'd expect some tradeoff there.
What makes this a great demo is that obviously the designers studied the physics of horses walking, trotting, and galloping. The physics have actually been well known for some time now, by equestrians if no one else. And the designers applied these same techniques to a machine. The result is something that to a human appears "familiar" for a living creature, but not familiar movement from a machine. Hence the wonder.
I will like to see what the military uses. I know they must have cracked some of these systems but leave them classified. From Stanford to CMU, this has been going on for decades and I can assume, the Army has a really better one which no one has seen.
Re. the gasoline-fired engine: Energy Density. And the need to drive a hydraulic (or pneumatic) pump directly.
These sort of machines got their start in the MIT Walking Machines lab decades ago (I know--I applied). But back then, they were tethered for both mechanical power AND computational power. Most people then couldn't comprehend that a gas-powered engine would solve the mechanical limitations, and time would solve the computational bottleneck. (I could--but I wasn't hired!)
But, they've grown into some pretty amazing devices! I'd hate to get on their bad side, though....
I have seen the videos of the big dog and I thouht it was quite impressive, but this Wildca package is even more impressive. I thought that I saw a number of different gaits, so it may be able to handle a lot of different situations. But it would never be battery powered at that speed. I can see quite a few applications for a "beast" like it, both law enforcement and military. Just picture this beast charging at the bad guys in a standoff situation. It could easily have armor added and travel on a path making it hard to target. And it could simply bash into somebody and allow the humans to follow, sort of like a blocker in football. In fact, can you imagine teams of these playing football? THAT would be quite a game, and probably cheaper than the pro football players today. How is that for an unaticipated application? They would need to look a bit nicer, though. It could be the first real "Iron Man" football game.
>> Consider the possibility of two teams of these robots playing pro football. Such an interesting possibility.
That exists actually. There is the Robo World Cup (soccer though). It evaluates how teams can work together to build better robots collaboratively. They use sports to test how efficiently the robots can work in teams if deployed in some challenging environments.
goafrit, I was not aware that it had actually been tried. Another really interesting concept would be to just put two robots on each team. Quite possibly a robot might break when sacked, but possibly not. It would be interesting no matter what.
No no - the football here is not American football. It was tried in soccer which is global FOOTBALL. I am not sure how you can do this with American football with the sacking. The soccer one looks at the collaborative part of the project.
goafrit, that is my point, sort of. American Football, which is more violent than soccer, at least during the games and between players. Soccer is strategy, skill, and speed, while football adds some serious impacts, actual tackles, and blocking. That is where a robot of some form could have a serious advantage, possibly with the option of leaping over an opposing player, which would be awsome, no doubt, or else just continuing to run after being tackled. And a robot able to take the ball and run a touchdown just a bit faster than those chasing might have a whole stadium standing for the play. It would be awsome to see that "big dog" robot grab a ball and sprint 90 yards for a touchdown, even more if the opposing team was just inches behind. I am not sure if there is a similar play in soccer.
Interesting - it alll depends if there are testing ruggedness at this phase. My understanding is that the focus is the brain power of these robots which can be examined from the soccer element. How rugged the robots are may not be that useful. You can make robots tough, that could seem easier, compared to making them intelligient.
You are certainly correct, at least it seems that way, that making robots tough should be easier than making them really smart. But sometimes both are needed, either at different times or sometimes all at the same time. Just like people, they need to be able to handle exceptions, both physical and those requiring a rapid analysis of data.
I'm impressed by several features in this demonstration. First, the robot "gets up" from a "reclining" position, secondly it manages to move at a variety of speeds, and finally it manages to recover from a "fall". A great demonstration. Next I'd be interested to see how well it can maneuver around obstacles, how well it manages on rough terrain, and how it climbs stairs. Finally, an aesthetic consideration ... when that "insect buzz engine" is muffled and the pitch dropped, it will become much more welcome in human company. In that regards, animals have it beat many times over as they quietly graze or scamper in a field.
I once commented to a bunch of robot builders about the importance of the fast-cycle, no-holds-barred Skunkworks methodology for getting innovative new systems up and running. The Boston Dynamics guy there instantly chimed in that they are true believers in the Skunkworks methodology for doing exploratory (pre-hardening) designs. The proof is in the pudding, or in this case, in the galloping robots! BD really does do great work, even though watching their robots in action scares the bejeebers out of me sometimes... :)
KB3001, Making a robotic anything internet connected is an invitation for hackers to do damage. So really, think about it: an internet connected robot suddenly under the control of somebody who intends to do damage. That is a just plain BAD IDEA. Even much worse with a fast moving high powered robotic animal.
Unfortunately we do have a buch of folks who are unable and unwilling to admit that providing computer controlled stuff with both intellect and self awareness could possibly have any potentially bad results. Just look at the windows operating system to see some of the bad results that come from a program having even a small amount of self awareness.
@WKetel, that's the usual argument between security and connectivity. Ultimately connectivity wins, otherwise we would have never had the internet :-)
When I said internet, I did not necessarily mean good old plain TCP/IP. It's got to be secure of course e.g. through bespoke protocols, encryption etc. The functional and cost gains of higher connectivity outweigh the security fears IMO.
Indeed, providing such powerful machines with connectivity is an invitation for trouble, but at the same time, it seems unthinkable that they would not be able to network either with each other or at least with their human controllers.
"Unfortunately we do have a buch of folks who are unable and unwilling to admit that providing computer controlled stuff with both intellect and self awareness could possibly have any potentially bad results."
That almost sounds like a reference to the Terminator movies, and indeed some of these advanced robots make it seem that not everything in those movies was so far-fetched. We don't need to extend our imaginations all the way to humanoid cyborgs with silicon brains comparable to a human brain -- machines such as successors to the Wildcat, outfitted with weapons and sophisticated AI would be cause enough to have real concerns about network security and hacker-proofing.
AZ, I was not referencing the terminator movies at all, just the fact that a lot of people simply are not aware of the secondary results of things that may seem like a good idea until they are instituted. Then the additional and secondary effects start happening and suddenly it is a lot like speeding up and passing everybody else on the road, then realizing that they were slowing a bit for a sharp turn. Of course there could be some benefits to some form of connection, but not through the internet, and not using internet protocols. In some cases it is much better to not have everybody on the same bus. In addition there is no benefit to have such systems accessible to the internet, except for some short term convenience. And just like most things done exclusively for short term convenience it would invite a future disaster. That is why you lock your doors.
Thanks for the clarification. Nevertheless, intelligent robots like the Wildcat do evoke concerns about maintaining control -- security concerns -- that are suggestive of a Terminator-like "machines out of control."
@AZ, this is one of those instances where it turns out that fiction had a fairly good guess at future reality. But I hope the rest of that plot remains in the realm of fiction. I could do very well to never need to deal with time-traveling robots of any kind.
Join our online Radio Show on Friday 11th July starting at 2:00pm Eastern, when EETimes editor of all things fun and interesting, Max Maxfield, and embedded systems expert, Jack Ganssle, will debate as to just what is, and is not, and embedded system.