Go straight to the Walking Robot Page (C.O.M.R.)
I have had had a long-standing interest in robotics and Protosynthetics. My reasons are I wish to construct better man machine environment interfaces.
This is with a view to make extreme condition Mecha (exoskeletons) for use in rescue operations in event of natural or other disasters. Also it is with a view to make artificial parts for people with disabilities, such as limbs etc. However, I am interested in non-invasive techniques, hence the focus on exoskeletons. Concider for a moment that a working exoskeleton, could ofer a disabled person mobiltiy, as well as functioning as a normal atonomus robot when required. FFurther It would lend itself to training by a human.
Further being able to make the prosthetic completely removable, i.e. no implants etc is an advantage with repect to contamination of the host. If an implant scenario were to be followed I would rather do it in a biogenetic method, where possible, i construction from with in by nanite, or cellular technologys.
To this end I have completed an honors degree in Mechanical Engineering. As my final project I designed and built a small waking (climbing) robot.
Exoskeletons and Telechairs. These terms refer to electro/mechanical interfaces that humans use and control. You could say that a car is an exoskeleton or telechair. That is it is an interface for a human being that allows an amplified interaction with the environment. Currently there is still no computer that can drive a car as well as a human, or even remotely close.
I believe that we should make robots to be human operated, or at least trained. Like the constructs of Mosher, the Hardi man or Walking horse. While these have their drawbacks, the human brain is excellent in accommodating imperfect mechanical systems. Just look at how primitive the actual control over a car is, in comparison to the control your own body offers you. After practice you can use a wide variety of cars and still drive it with competence after a very short time, and hold a conversation at the same time. The control systems could be added on as a layer over this.
With particular consideration of walking robots I believe that the main problems with most designs that I have seen is that they usually try for a static stability by using an insect or octopod arrangements. There are notable exceptions such as MIT's effort with all types of hoppers and other multi pods which are most impressive.
As far as walking robots or almost any kind of classic animal or humanoid autonomuos typerobot designs so far, there a two areas that they usually deficient in.
1) Mechanics. The design of the componets are almost always, ungainly. I'm sure the designers hope that the contol system can overcome the inherent back lash, poor tollrences, poor weight to strength ratios and general blockyness of the robots. Many bipedal robot simply have a big rectangular block for a torso, and then use and advanced inverted pendulim control sytem to hopefully compremize. Consider for a moment the complexity of the muscle arangement and the subtletly of movement you see in the animal world in the torso region. Where you see a lower complexity you have arthropdia, but look at the scale they exist on. This assumption carries over into the gait and static stabillity concepts of a robot. To solve this problem, a little, robots should be designed soth that components have 'grace' i movement and integration. For example. Most robot I belive should be hydraulic. The channels/hoses for the fluid could acttully be inlayed, and flow through joint seals. This eliminates much of the imbalance of motor drives etc, and gives a good power to eight ratio. Further there should be no central power system, but a series of self contained motor, eletric, and fuel type though out the robot, almost one per major movement group. This allow the robot or exo skeleton to continue functioning even if it loses limbs, muscle sections or power to varius part. also the over all fluid systems becoms lees complex, as there needs to be no major interface manifolds. Even the joint and 'bones/exoskeletons' need to be considered for their elastic and dampening properties. I relize the issue here is cost. I am talking about possibly a very expensive robote design. Although mybe not. It depends. After some consideration, $300000US should be enough to get the mechanics workinig. It depends largely if you have to design the acuator sytems or you can by siome of the shelf.
2) The implimentation. Why do so many walking robot designers becom fixated on walking robots, I see that a robot needs to be able to deliver a payload from point A to point B in the quikest possible time while takin up the smallest envolope of space, and consuming small amount of energy. I think that wheels and tracks have a play in all land based locomotion. Consider for amoment the places in a human environment where wheels may be used, Therer are enough to make them a viable option. Even in a no human enviroment. Thus I belive wheel should be present in the robot system in some way. I have yet to see a robot philosophy, that divers awya form, fixed body movable legs. Ie we wil create abody, like an anthrapod, and then attach movable legs. The assuption is that insect are statically stable. it seems though as pointed out by Binnard, that cockroaches are actually dynamically stable at medium to hig speed. It may be possible thatthere is no real distinction between a dynamically stable system and a statically stable system. The time fame only varies. For example even in the most statically stable system, there will be times when the legs that are in the air reduce the static stability even slightly due to the centre of mass of the robot no longer being coincedent with the desired geometric centre for the the most stable configuration. In a dynamically, it is in a way alway stable, statically, and it may be argued that infact is more statically stable that a statically stable design, as at all points in time the system is moving in such a way that the actual centre of mas is very close the desired geometric centre of mass. This said I belive the body of the robot should be able to move relative to the legs of the robot. The results of this although applied some what to an extreme may be seen in the COMR design.
I will hpefully inclue a short movie of the COMR that I made from technical lego. It was able to climb to a height of about 20 cm, which is about 1.5 times its total hieght. It would be possible to extend this to a factor of 1:5 without significan problems, thus a meter step may be made. To date few robot if any can achive this ratio of step to body hieght (note the definition of body height is given as the normal frunnin orientation, it may be agrued that this is biased in the COMR's favour), again, MIT hasn a sumersaulting biped and a mopod (I hightly recomend you see the clip), they could probally out perform this if they desired. The one draw back with the mit robots, is, I imagine the tether. It seems, although I am not ceratin that they require to be tetherd, which is not really as bad as it seems in some circumstances.
A solution to the problem is given here: (more later)
COMR, A WALKING ROBOT