by StuartL » Sun May 11, 2008 3:08 pm
by StuartL
Sun May 11, 2008 3:08 pm
In your case you can probably do what most people do with their robots, and cheat.
Most of the modern domestic robots (i.e. Robo-1 style robots) have a pre-coded sequence of moves which are effectively just patterns causing the robot to do things. The moves take no account of robot orientation, inertia etc other than as coded originally.
This means that provided the robot is in the position you expect it to be when you trigger the move sequence everything works out. All (most) of the cartwheels you've seen use a precoded sequence of servo positions to throw the robots balance over its "hands".
The disadvantage to this approach is the consequence when the unpredictable happens. Say there's an uneven surface and when the robot attempts to take a step forward the balance is thrown slightly forwards. It may not cause a problem immediately but the robot could fall forwards when it attempts to complete the step. Similar problems occur when the robot makes contact with another robot.
The solution to this is to program the robot to react to its own environment, which is what you were getting at in your original post. This is possible but it's VERY hard to do. And, on top of that, it's almost impossible to make a robot react quickly on a domestic budget.
If you have the time, inclination, money and patience you could make a robot or two have reactive capoeira fights. I suspect that, given your original post, you'd be better off getting the kits and spending the time programming your capoeira moves to be as stable as possible on a variety of surfaces and make them human controlled for your project. That way a human can react to them falling over and press the button to make them stand up again.
Your selection criteria would therefore be:
1) Wireless control.
2) Move sequence reprogrammability.
3) Easy programming of stances, poses and movements.
In your case you can probably do what most people do with their robots, and cheat.
Most of the modern domestic robots (i.e. Robo-1 style robots) have a pre-coded sequence of moves which are effectively just patterns causing the robot to do things. The moves take no account of robot orientation, inertia etc other than as coded originally.
This means that provided the robot is in the position you expect it to be when you trigger the move sequence everything works out. All (most) of the cartwheels you've seen use a precoded sequence of servo positions to throw the robots balance over its "hands".
The disadvantage to this approach is the consequence when the unpredictable happens. Say there's an uneven surface and when the robot attempts to take a step forward the balance is thrown slightly forwards. It may not cause a problem immediately but the robot could fall forwards when it attempts to complete the step. Similar problems occur when the robot makes contact with another robot.
The solution to this is to program the robot to react to its own environment, which is what you were getting at in your original post. This is possible but it's VERY hard to do. And, on top of that, it's almost impossible to make a robot react quickly on a domestic budget.
If you have the time, inclination, money and patience you could make a robot or two have reactive capoeira fights. I suspect that, given your original post, you'd be better off getting the kits and spending the time programming your capoeira moves to be as stable as possible on a variety of surfaces and make them human controlled for your project. That way a human can react to them falling over and press the button to make them stand up again.
Your selection criteria would therefore be:
1) Wireless control.
2) Move sequence reprogrammability.
3) Easy programming of stances, poses and movements.