Shadow Biped Walker by David Buckley
- Shadow Walker - Retired
- A two legged human sized pneumatic powered
walking robot.
Design and building started 1988.
Size - 5ft 6inches high, weight about 80lb
Operational area:-
safety supporting frame 4ft * 8ft * 8ft
plus host computer, a small compressor fits
under the host computer's table.
Back in 1988 when the Shadow Group was new, Richard Greenhill, the founder and funder of the Group, was working on a shrugging shoulder which he thought was necessary for a robot or human to lift heavy weights such as suitcases. For me this wasn't a very fascinating aspect of robotics and I suggested that in order to attract members we needed an appealing and challenging project such as the full sized biped robot we had often talked about. Rather than wait until everything was developed we could start with the legs and add the other parts as we went on. So the project to build the first Shadow Biped Walker was set in motion.
I drew up plans for the skeleton, researched medical texts for muscle placements and designed the electronics so we could run the walker from an Acorn Archimedes computer.
Richard Moyle and myself built the skeleton and the electronics were built by various other members of the group. Richard Greenhill designed and built the muscles (his development of the McKibben muscle) and we attached them to the legs. All the joints had potentiometers attached for position feedback and all the muscles had tension sensors. The next step was to build the pneumatic valve sets to drive the muscles.
I was working abroad and when I came back others in the group had built the two sets of air valves and switches seen in the photograph. They took up all of the body space leaving no room for any arm mechanism at all! I was very disappointed but what was built was built, maybe it was my fault for not drawing up the arms. David Tricket built a pressure sensor panel with modified pressure dial gauges so we and the computer could read the pressure in each muscle and this was attached to the back of the robot.
Richard Walker took over writing software experimenting with all sorts of neural nets trying to get the walker to learn how to stand.
Since the robot used McKibben muscles I hoped it would be quite flexible but Richard Greenhill kept pressing for more force from the muscles and in the end it was very highly strung and could nearly stand up without any compressed air in the muscles at all.
During this time I was spending time working abroad so had little control of what went on with the robot and one time when I returned The group had decided that since some of the potentiometer fixings had come loose instead of securing them they would take off all of them and replace them with analog optical sensors.
After a lot of work especially by Richard Greenhill, in getting the sensors to work, Richard Walker discovered they were not monotonic and so more time and effort was expended on them.
The original valves were 110vac (used because Richard Greenhill bought them cheaply from Proops at I think £1 each) but Richard Walker thought they didn't have enough through flow to fill the muscles fast enough for his control software to work. So we changed them all for 12v washing machine valves with enormous flow capacity. In doing so we discovered that the reason the old valves couldn't supply the necessary air was that most of the connections in the pressure sensor panel virtually sealed off the supply and simply making them good would have been sufficient. The new valves required large fittings and only cast-iron ones were available, this set another problem because the group member who modified them didn't clean them thoroughly and they all went rusty and particles stopped the vales from working. An all night session stripping and cleaning everything was required so we could show the robot, probably at the Robot Olympics 1996.
The new valves brought their own problems, they had rubber diaphragms and pressure bleed holes and didn't respond well to being pulsed,but Richard Walker eventually managed to tame them.
In 1998 I moved into a new house in the North of England and the walker was transferred from the Shadow labs in London with the idea that I could continue to work on it at home. However after inspecting it I realised that it would take as much work to refurbish as to build a new walker and so it was decided to retire it. The walker now hangs on the wall of the Shadow labs.
For more information see http://www.shadow.org.uk/projects/biped.shtml
Many sites have refered to the Shadow Biped, NASA's Robonaut pages used to have a survey of Humanoid Robots around the World [http://vesuvius.jsc.nasa.gov/er_er/html/robonaut/Other_bots.htm] which included the Shadow Biped but that page has now gone (the text
though is on the Wayback Machine), The best way of avoiding broken links is to ask Google.
A good place for Biped links is androidworld.com - see Biped Projects
[photographs of the Shadow Biped by Richard Greenhill]
