Robots Take To The Stairs – This Is Just The Beginning

113 9 Loading

Robots can climb stairs, and they are doing it everywhere you look.  “No big deal” you say, but it really is a big deal.  Five to ten years ago almost nobody was doing it.  Now grad students are doing it all by themselves for thesis projects.

Stair climbing may not seem like it is that hard, but it really is hard, especially if you don’t take the easy way out by using a tracked system.  Stair climbing only looks easy because we are building on the knowledge that those before us have so painstakingly accumulated.  This is a common theme with technological advancement: things that once seemed hard (and really were hard!) often appear easy once we figure out how to do them and they become commonplace.

One of our favorite stair climbing robots is the appropriately named stairbot.  Not only is the stairbot pretty awesome, but it is accompanied by an excellent website that offers a nice look at the engineering theory behind stair climbing robots.  The stairbot can drive back and forth, pivot and turn, drive over small obstacles up to 4 cm high, and yes,  go up and down stairs.  Lets take a look at the stairbot in action:

One of the techniques that is key to the stairbot’s success is its spindle driven ability to dynamically change its length.  In order to navigate stairs both upwards and downwards, the stairbot is equiped with several sensors that allow it to sense its orientation and then make the necessary adjustments.

Below is a frontal view of the stairbot in all of its beauty, sensors and all!

stairbot_frontal

Perhaps the easiest and most common technique for getting robots up stairs it to equip them with tracks, similar to your typical army tank.  The tracked technique may require less skill and precision than other techniques, but still it gets the job done.  Below is a video of the B2P2 robot, an Unmanned Ground Vehicle (UVG) initiative developed at the University of Angers in France.  Also below is a video of the versatile army robot from iRobot, called the Packbot, which also uses tracks.  There are several other tracked robots that go up stairs – these are only two such examples:

Of course one of the harder challenges in stair climbing is the humanoid variety.  The famous Asimo robot from Honda conquered the task many years ago, as did the HRP-2 line of robots.  Lets look at videos of the Asimo robot, first succeeding at stair climbing, and then in the second video failing.  The video with the robot falling is quite hilarious so laugh all you want…but keep in mind that robots are likely to have the last laugh!

Finally, we end our survey of stair climbing robots with a short video of a robot developed at the department of Mechanical Engineering and Mechatronics, Ariel University Center, Israel.  We don’t have much information on the robot, but you have to admit the climbing technique is quite interesting and novel:

Stair climbing, a difficult challenge by itself, is just one of a long series of major challenges that robots can now perform quite easily, yet nobody seems to notice.  Go ask the general public about the progress robots have made in the last ten years and most people will say “not much”.  Even though we are bombarded by small, specific signs of progress everyday, it is hard for us to step back and see the big picture.

Take a moment to step back, however, and the robotic revolution is all around us!  Robots are in fact advancing at an incredible pace, and they are not simply replicating human abilities – they are building upon them and extending them.  Humans can only climb stairs using their feet.  Robots can climb stairs using feet, but they can also do it with tracks, wheels, shifting appendages, and more.

Moving beyond stairs for a moment, robots like Big Dog show us that robots can handle just about any terrain imaginable.  The future certainly looks bright for robots.  Lets hope that translates into an equally bright future for humans too.

Discussion — 9 Responses

  • Guest May 6, 2009 on 9:09 pm

    I fail to understand how you can call this much of an “advance in robotics”. It is not much more than a bit of refinement of some mechanical engineering. It might be justified to say that “we now have small machines that can climb stairs much better”, but that is hardly a big robotics breakthrough.

    Let us make a distinction between incremental improvements and breakthroughs. Dean Kamen got approval to sell his functional wheelchairs that could climb stairs very well, 6 years ago. he first built them eariler than that. Further, the electronic sophistication of the device is greater than anything shown here. It was larger, yes, but it can fairly be said that it represents a higher level of “robotics” than any of the machines mentioned in this article, with the possible exception of “big dog”. And that was years ago, folks!

    Processing speed and power have advanced at a great pace. But a fast dumb computer is still a dumb computer. It can simulate “intelligence” a bit better, by virtue (and cost) of the vastly more complex software that it can now run because of that speed. But again, these are incremental improvements.

    There have been no real breakthroughs either mechanically or computationally in the field of robotics for decades. Until one does occur, do not expect anything resembling true “AI” or independence, in the foreseeable future. (Breakthroughs, by their very nature, are not foreseeable.)

    I like your website, but things need to be kept in perspective. “AI” has been relatively stagnant for a long time. The changes we have been seeing have been incremental: evolutionary rather than revolutionary. It is rather presumptuous to predict major advances when all we have actually seen are incremental changes.

  • Lonny Eachus May 6, 2009 on 5:09 pm

    I fail to understand how you can call this much of an “advance in robotics”. It is not much more than a bit of refinement of some mechanical engineering. It might be justified to say that “we now have small machines that can climb stairs much better”, but that is hardly a big robotics breakthrough.

    Let us make a distinction between incremental improvements and breakthroughs. Dean Kamen got approval to sell his functional wheelchairs that could climb stairs very well, 6 years ago. he first built them eariler than that. Further, the electronic sophistication of the device is greater than anything shown here. It was larger, yes, but it can fairly be said that it represents a higher level of “robotics” than any of the machines mentioned in this article, with the possible exception of “big dog”. And that was years ago, folks!

    Processing speed and power have advanced at a great pace. But a fast dumb computer is still a dumb computer. It can simulate “intelligence” a bit better, by virtue (and cost) of the vastly more complex software that it can now run because of that speed. But again, these are incremental improvements.

    There have been no real breakthroughs either mechanically or computationally in the field of robotics for decades. Until one does occur, do not expect anything resembling true “AI” or independence, in the foreseeable future. (Breakthroughs, by their very nature, are not foreseeable.)

    I like your website, but things need to be kept in perspective. “AI” has been relatively stagnant for a long time. The changes we have been seeing have been incremental: evolutionary rather than revolutionary. It is rather presumptuous to predict major advances when all we have actually seen are incremental changes.

  • Keith Kleiner May 6, 2009 on 9:20 pm

    Lonny,

    I respectfully, but strongly, disagree with your comments.

    First of all, the post does not claim that these robots represent major breakthroughs, nor that they require the resources and ability of a Dean Kamen to create. In fact, that is exactly the point! 6 years ago we needed Dean Kamen to make stair climbing robots that worked well, now people are creating robots that can do this all over the world, in several different ways, some more sophisticated than others, but all functional. This represents a “sea change” in robotics that I think you are failing to see.

    Robots today are doing surgery, serving drinks, performing synchronized dancing, and yes, even climbing stairs today. These achievements were either non-existent or found only in the realms of the Dean Kamen’s only 5-10 years ago. Now individuals can make robots that do these things in their garages with limited resources. How can you not recognize that this is an amazing advancement?

    The advance of robots is not represented by a single major breakthrough…it is instead the culmination of millions of incremental improvements that are happening on a daily basis. While people like you are asleep at the wheel claiming that robots are not improving, those who step back and look at the big picture know that the real story is that robots are improving and adapting with alarming speed.

    As a final comment, I will note that you are mixing AI and robotics interchangeably, as if they are the same thing, but they are not. Coincidentally, the same story that we are seeing in robotics is indeed occurring with AI. Most people, you included no doubt, will claim that AI has not made any advances in the last decade, but those who can see the bigger picture know otherwise.

  • Keith Kleiner May 6, 2009 on 5:20 pm

    Lonny,

    I respectfully, but strongly, disagree with your comments.

    First of all, the post does not claim that these robots represent major breakthroughs, nor that they require the resources and ability of a Dean Kamen to create. In fact, that is exactly the point! 6 years ago we needed Dean Kamen to make stair climbing robots that worked well, now people are creating robots that can do this all over the world, in several different ways, some more sophisticated than others, but all functional. This represents a “sea change” in robotics that I think you are failing to see.

    Robots today are doing surgery, serving drinks, performing synchronized dancing, and yes, even climbing stairs today. These achievements were either non-existent or found only in the realms of the Dean Kamen’s only 5-10 years ago. Now individuals can make robots that do these things in their garages with limited resources. How can you not recognize that this is an amazing advancement?

    The advance of robots is not represented by a single major breakthrough…it is instead the culmination of millions of incremental improvements that are happening on a daily basis. While people like you are asleep at the wheel claiming that robots are not improving, those who step back and look at the big picture know that the real story is that robots are improving and adapting with alarming speed.

    As a final comment, I will note that you are mixing AI and robotics interchangeably, as if they are the same thing, but they are not. Coincidentally, the same story that we are seeing in robotics is indeed occurring with AI. Most people, you included no doubt, will claim that AI has not made any advances in the last decade, but those who can see the bigger picture know otherwise.

  • Peter Sandilands May 6, 2009 on 6:44 pm

    I agree with Keith here and just want to give an example of excellent AI that is good enough that we don’t notice it at all: Google. Although it doesn’t emulate a human, it takes intelligence to wade through the pages of the internet and index them so as when someone searches for a query, in less than a second it can come back and give you almost always relevant information.

    This is also an example of lots of small improvements being unoticable by the general public, most would be hard push to tell you what changes Google has made in the last 5-10 years, yet there has been constant development going on improving the backend. Things are progressing and this makes me happy.

  • Gabriel May 7, 2009 on 8:06 am

    For fast stair climbing you must see RHex:
    http://kodlab.seas.upenn.edu/RHex/Home

    I’ll try to find a video online.

  • Gabriel May 7, 2009 on 8:08 am

    Ok, I found and online video of rhex climbing stairs:

    http://sandboxinnovations.com/index.php?leaf=12

    Best.

  • Howard Barnes September 29, 2009 on 4:55 pm

    It does solve the task of climbing stairs but only by virtue of dedicating its entire construction, no? If the principle which it embodies is abstracted further and condensed into a modular unit, perhaps it could then be integrated with a fully-functional robot?

  • Howard Barnes September 29, 2009 on 12:55 pm

    It does solve the task of climbing stairs but only by virtue of dedicating its entire construction, no? If the principle which it embodies is abstracted further and condensed into a modular unit, perhaps it could then be integrated with a fully-functional robot?