I'm having some trouble mentally figuring out how ED-209 could be combined with a two-wheeled PID balancing robot. In a two-wheeled balancing bot, the gyroscope is ideally placed close to the axis of the two motors. If ED-209 were to have "PID feet", would the gyroscopes go in his feet? or would one gyroscope go in his torso to keep it balanced while his feet walked here and there?
And if anyone can also throw in their "2 cents" about biped robot walking that would be appreciated.
I've done some preliminary research on biped walking/running and it seems to be ultra difficult, but its been achieved in commercial products like Kondo robots, Bioloid, and even the Lynxmotion Biped scout.
Adding wheeled-feet to a biped just seems over-the-top complicated, but its still interesting to ponder about the challenges that one has to face if attempting to build such a mechanism.
What is ED-209? There are two types of robot walking. The first is static walking, where the robot’s COG is always over one foot or the other. Then there is dynamic walking, like people do, which is closer to a controlled fall in that the COG (center of gravity) is, at times, not centered on either foot or any combination of feet.
Static walking is used by almost all walking robots and is not difficult to do (to get it to look natural is more difficult). Dynamic walking requires faster motors and more computation than most hobby robots have; the programming problems are also much greater.
Balancing robots usually depend on having two wheels in a specific layout. In all balancing robots I’ve seen they have two wheels side to side in fixed locations. Making a balancing robot with movable wheeled feet would be difficult.
I would suggest starting by building a traditional balancing robot. And again, there are different levels of balancing robots. The best I’ve seen on YouTube can be found with the search “balancing robot iced tea” because it is used to serve iced tea in the video.
Most use both a gyroscope and an accelerometer. An MPU 6050 should do. The major constraints on the position of the sensor is that it should be centered between the wheels and also directly over the COG on the x and y. The z can be anywhere.
It really depends on what you want to do exactly, with wheels I think there’s two possibilities:
Normal balancing robot: two wheels next to eachother, and importantly, no other degrees of freedom, so no ‘knees’ for example. Easy to do, just like other balancing robots, place of accelero shouldnt matter so much because the angle to the earth stays the same (only you have less ‘noise’ from swinging movement when accelero is close to feet, but then again, robot shouldnt swing so much anyway and also you could take it into account in calculations).
Same, but with extra degrees of freedom, eg in the knees. There’s two sides to this: with wide wheels and when the axes of the wheels do not coincide, you can have a somewhat stable position, maybe even emulate some walking movement. However, because of the knees the control becomes more involved, especially when not in a stable position. I dont know if somebody did something like this already, but likely it would require you to do a lot of experimenting. Preferably your actuators would give feedback on their position, and if thats accurate, 1 accelero is theoretically enough. Certainly a challenging project!
Without wheels may be possible as well, Im working on a small pedal robot myself that kind of resembles this robot with some imagination. Dont know if it will work out yet
thanks for the breakdown and detailed information. I’m sure your pedal bot is going to be revolutionary and inspire others to create bots of all kinds!!!
