A.I.
A.I.
by sktbrd627
Wed Jan 03, 2007 10:25 am
Hi, I'm new to thsi forum and was thinking about buying a bioloid. I had a question about it, though. I was curious to see if anyone has worked towards developing an A.I. for it? I guess i made the mistake in assuming yes since you can program it in C. I was just curious though. Any info is greatly appreciated 
. Thanks
. Thanks
by limor
Wed Jan 03, 2007 1:01 pm
It depends one what you regard as A.I.
The way i see it there are 2 levels of A.I.
1) Path planning, problem solving and "high-level" intelligence. You will find many posts on this forum from people who got their robot to plan its objectives and then update it in reaction to sampled sensor data. Sensors can be the AX-S1 of the Bioloid, a camera or any other variety of sensors. This path planning is what is commonly referred to as A.I. in video games and robot competitions.
2) Another form of A.I. is a lower level of control and path planning which is a bit more complicated and allows the robot to be less prone to falling and their motions to be more aesthetic like humans/animals. Check out Neural-Control, Fuzzy-Control, P.I.D, Kalman-Filter, Inverse-Kinetics etc. Honda's Asimo has surprizingly human motions also because it deploys advanced of control. RC-Servo control and most hobby robot controller-boards use an open-loop control scheme so the motion is pre-defined and it is executed no matter what goes on in the environment. To acheive better motion control, the motors have to continously be updated in reaction to sampled sensor data. The way hobby robotics gyros work is a good example of real-time motion update in reaction to sensory input.
Once advantage of the Bioloid servos is that they do allow you to perform your own low-level control.
The way i see it there are 2 levels of A.I.
1) Path planning, problem solving and "high-level" intelligence. You will find many posts on this forum from people who got their robot to plan its objectives and then update it in reaction to sampled sensor data. Sensors can be the AX-S1 of the Bioloid, a camera or any other variety of sensors. This path planning is what is commonly referred to as A.I. in video games and robot competitions.
2) Another form of A.I. is a lower level of control and path planning which is a bit more complicated and allows the robot to be less prone to falling and their motions to be more aesthetic like humans/animals. Check out Neural-Control, Fuzzy-Control, P.I.D, Kalman-Filter, Inverse-Kinetics etc. Honda's Asimo has surprizingly human motions also because it deploys advanced of control. RC-Servo control and most hobby robot controller-boards use an open-loop control scheme so the motion is pre-defined and it is executed no matter what goes on in the environment. To acheive better motion control, the motors have to continously be updated in reaction to sampled sensor data. The way hobby robotics gyros work is a good example of real-time motion update in reaction to sensory input.
Once advantage of the Bioloid servos is that they do allow you to perform your own low-level control.
by sktbrd627
Wed Jan 03, 2007 10:59 pm
thanks limor! I think i would like to work towards your first definition. I want it to be able to learn from it's environment, and act accordingly. Although, I definitely want it walking more human-like, but at this time I don't even know how I will build it. I was really interested in Jon's wiki for his MicroRaptor...That's more along the lines of something I'd like to do. I want to create, or attempt to create, something that can learn/remember each time it is turned on. Eventually, I'd like to move onto more complex things, such as having it be able to think and solve problems it encounters, but for now I'm just focused on very basic level A.I. I don't entirely think this is out of reach, the hard part will be me programming it, and finding out a way to have it walk human-like without falling. Thanks again you guys, and Jon I will be keeping a close eye on your MicroRaptor. I'm very interested in it.
Ali
Ali