Calculating Torque required to self Right

[AdamHeard]

I was wondering how to calculate how much torque is required to self right using a single jointed arm. Right now, I'm just doing weight of the robot (~150) at the end of the lever arm to calculate the torque. In reality, less would be required because you are not lifting the entire robot weight (right?).

Anyone know the exact way to calculate this?

[Qbranch]

You need to do a summation of torques. Using the side of your bumper/robot that you flipped over as an axis of rotation, figure out how much torque the robot provides on the right trying to flip the robot over, then figure out how much torque is holding the robot down on the flipped side. Subtracting the robot's torque from the flipped side's torque will give you how much torque there is holding the robot down. Then, you can figure out how hard you need to push to flip the robot back rightside up.

Of course, this'll only be a close estimate still, since as the robot rotates back to the upright position (or depending on how far it flipped over) your component of the force of gravity pushing down on your robot will increase and decrease as the angle changes relative to the ground, since as the robot rotates upright more of the force of gravity will be counteracted by the wheels on your robot.

My question is... wouldn't it be better to design the robot NOT to flip easily? Then again, I think you may be on to something this year... I'm thinknig there's going to be a lot of robots crashing into those operator stations as people try to take those turns too fast.

*makes mental note to definitely add accelerometer G-limiter software into teleoperated control loop*

-q

[MrForbes]

I would start by caculating the exact center of mass of the robot, then figuring out how far that is located from the edge of the robot where it will be resting as the righting action begins. Do a sum of the moments about one point, then you can figure out the force required to push the robot back upright. This will change as the robot moves, of course...so you probably need to evaluate it in several positions (or figure out a nice mathematical model of the whole operation, which sounds like too much work to me)

[AdamHeard]

Originally Posted by Qbranch

My question is... wouldn't it be better to design the robot NOT to flip easily? Then again, I think you may be on to something this year... I'm thinknig there's going to be a lot of robots crashing into those operator stations as people try to take those turns too fast.

*makes mental note to definitely add accelerometer G-limiter software into teleoperated control loop*

-q

We have a pretty low CG and don't plan on tipping; but with the high speeds and the overpass that could easily close line robots, I'd like to know we can get back up.