Spring Flamingo Frequently Asked Questions

Last updated 1/12/2000, Jerry Pratt

What materials did you use for Spring Flamingo and why?  Spring Flamingo's body parts are made mostly of aluminum and carbon fiber.  Aluminum is used because it is light weight and easy to machine.  Carbon fiber is used because it is light weight.  Carbon fiber is hard to machine, though and very expensive to make custom parts.  So all the carbon fiber in Spring Flamingo is in carbon fiber tubes.  The legs are carbon fiber tubes with aluminum plugs glued in the ends.  The body is all aluminum frame.  Most of the standard purchased parts such as shafts and bearings are made of steel.

What are the advantages of this robot?  The advantage of legged robots over wheeled robots is that legged robots can walk in many places where wheeled vehicles can't go such as through forests, over holes, up stairs, etc.  The advantage of Spring Flamingo over other bipedal walking robots is that it has force controlled actuators, rather than position controlled actuators.  This makes it easier to coordinate the joints of the robot and to control the ground reaction forces on the feet.

What are the disadvantages of Spring Flamingo?  The major disadvantage of Spring Flamingo is that it is tethered and attached to a boom.  This means it can't go anywhere except around in a circle.  Also, since it is only planar, we can't use it to study the side to side aspects of walking which are very important in human walking.  Also the robot doesn't have any vision, or audio sensing.  Right now it is used only to study locomotion.

What motions can the robot perform?  Spring Flamingo can stand and balance, walk slowly and quickly (1.1 m/s to speed), and stop.  The robot is attached to a boom which makes it planar.  It can move forward/backward, up/down, and rotate.  It cannot move or rock side to side though.

What are the uses of Spring Flamingo?  Spring Flamingo is a research robot used to study bipedal walking.  The robot itself is not useful for doing work, mostly because it is only planar.  However, the knowledge gained from Spring Flamingo will be used in future robots which may be able to do useful work.

What is the most difficult part of making Spring Flamingo work?  Getting to robot to walk was fairly easy.  However, getting the robot to walk quickly and gracefully was much more difficult.  Gracefulness requires coordination of the joints, and soft landing of the swing leg.  Quick walking requires lifting up, swinging, and placing the swing leg really fast.  A human does all this in about 0.3 seconds.  Getting the robot to swing its swing leg fast so that it could walk fast was probably the most difficult part of the project.

General Robotics Frequently Asked Questions Asked of Jerry Pratt

How does robotics affect our normal lives today?  Most peoples daily lives aren't affected at all by robots right now.  There are many automatic devices that we use to make living easier such as dish washers, answering machines, etc. but none of these resemble what's typically called a robot -- something that can move around and manipulate the world.  There are many machines which people use that require a human operator, such as back hoes, which would be considered robots if they acted on their own without requiring a human operator.  However, it will probably be a long time before such machines are made autonomous, if they ever are at all.  Some special purpose robots are used in industries such as spray painting and spot welding cars and assembling circuit boards.  However, none of these robots move around in human living environments and all of them perform specific pre-programmed roles.

In the next 10 years how are robots going to change are lives and why?  They won't affect your daily live much at all really.  There may be some simple robotic devices available such as robotic lawn mowers but nothing major like humanoid robotic personal assistants.  There are several reasons: cost, difficulty, and slow progress.  First robots are really expensive.  Why pay an extra $1000 for a robotic lawn mower when you can hire the neighborhood kid for $10?  Home robots with the same complexity of a human would cost tens of thousands of dollars even in large quantities.  Second, the tasks people would like personal robots to do such as laundry, get the newspaper, vacuum, etc. are very difficult tasks.  They only seem easy since humans are so good at doing unstructured things.  If a robot is going to do something like vacuum, it'll have to be a specific robot with wheels and the floor will have to be clear of junk like clothes and books.  Third, there won't be personal home robots in the next 10 years due to the slow progress of robotics research.  Robots today aren't much better than they were 10 years ago and they probably won't be much better 10 years from now.  The major reason for the slow progress is the difficulty of building robots and making them work, the small number of people working on robots, the lack of a commercial market, and the difficulty of sharing information between researchers since there aren't many good common platforms for doing the work (each group designs and builds its own robot which is different from the robots everyone else is using).  Contrast this with computers which are fairly simple well understood devices, tens of thousands of people are doing research on them,  there's a multi-billion dollar market for computers, and all computers are similar so what one researcher discovers for one computer probably applies to them all.  Thus it is no wonder computers have advances so rapidly while robots have good pretty much nowhere.

So in short you probably won't be affected by robots in the next 10 years except perhaps in entertainment.  Companies like Disney have used animatronics for quite some time now and will only continue to do so more.  Also, robot pets and toys such as Lego Mindstorm and the Sony dog Aibo will become more and more popular.  However, they won't do the chores for you.  They will be fun and educational though and definitely inspire tomorrow's robotics scientists.

Robots will get better though.  It's just going to take a long time.  So instead of asking how robots will change our lives in the next 10 years we should ask how robots will change our great-great-great grandchildren's lives 100 years from now.  For them maybe robots will be dumb personal assistants which can do things like follow you around and take notes and help carry heavy objects and play fetch or other things a pet can do.  But they most likely won't be able to think like a human and make good decisions.  For something like that -- a robot which thinks and feels and has intuition -- I'd say if it happens it'll be on the time scale of 1000 years from now.

What have been the most important recent discoveries in bipedal walking robotics?  There have been several recent discoveries.  One which was made with Spring Flamingo is that the natural dynamics of the robot can be used to simplify the control.  For instance, the swing leg can swing passively instead of requiring precise control to tell it how to move.  It was known since the 1980's that fully passive walking mechanisms could be built that walk downhill under the power of gravity.  However, this was one of the first times that those mechanisms were used to help a powered robot walk.  Another important discovery we made is how to make force controllable actuators.  We use a technique called Series Elastic Actuation which allows us to get a large range of forces from standard electric motors.  Another important discovery in robotics is how central pattern generators, which are found in most animals as neural oscillators, can be used in robot control.  Matt Williamson uses these in the control of the arms of Cog, a robotic humanoid.  Many people have used them in the control of bipedal walking simulations but only recently in real bipedal walking robots.

What do you think will be the next important discoveries in bipedal walking robotics?  The next important discoveries will be to find control strategies for 3D walking including dynamic turning, shuffling sideways, etc.  These will be found mostly by experimenting with the robots and trying out new ideas.  Down the road a new discovery will occur when biologist start to figure out how the control of human walking is wired.  This will give roboticists better ideas for control strategies.

Are you afraid of robots taking over the world once they are smarter than humans?  Absolutely not.  First, robots won't be as smart as humans for a long long time, if ever.  I'm talking hundreds to thousands of years.  Second, the question assumes that taking over the world has something to do with intelligence.  The smartest people don't necessarily run the world.  The most charismatic do.  Human intelligence is at such a level that being any smarter wouldn't necessarily help one become more powerful.  Finally, and most importantly, anything which makes a robot more intelligent will make humans more intelligent because they will use the technology themselves.  A person with a piece of paper and pencil is much more intelligent than a person without one.  Yet we don't fear paper and pencil taking over the world.  Likewise a person with a computer is much more intelligent than a person without one.  Artificial Intelligence technology will be used to improve portable digital assistants and wearable computers well before its used to make intelligent robots.  And finally a human with a personal cognizant robotic assistant will be smarter than a human without one and will be smarter than just the robot alone.  

Of course there will be profound implications if we make machines which are smarter than ourselves.  Just like there's profound implications of any other technology: cloning, nuclear power, etc.  And just like there will be profound implications when we find intelligent life elsewhere in the universe.  Should we fear all these things?  Probably not.  But we should understand them and how they affect us.

Copyright MIT Leg Laboratory. All Rights Reserved.