Impact force of a 6 DOF robot arm 2

[ddeck]

We have a 6 Joint, 6 DOF robot arm that has a gripper on the end for manipulating parts within a closed cell. The access door to the cell can be hit with the robot (since the cell is thinner than the working envolope of the arm).

I'm trying to size the hinges and latch that are needed on the door so the robot can not break down the door.

So, I'm trying to calculate what the maximum force the robot can generate. The two ways i see the robot hitting the door is by
-pushing it straight out (bench pressing it)
-or by swinging into it when all joints are at top speed.

The second mode is where I'm running into difficulties calculating the force generated on the door. Do I use Impulse, Momentum, Energy method???

Any ideas or opinions would be appreciated. Thanks

 

[GregLocock]

You've got two separate problems

1) determine the maximum speed with which the arm can hit the door

2) work out the forces generated by that impact.

1 is pretty straightforward, I assume.

For 2 you need to know the stiffness of the door, and the effective mass of the arm. Then you can work out the time of contat, which will give you the force.

The easiest way to solve all of this is a non linear multibody dynamics model, such as ADAMS, LSDyna, Nastran 4d (or whatever they call it). If you are going to do this sort of stuff very often it would be worth learning to do it in one of those programs.






Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[ddeck]

Any ideas on how to get the effective mass of the arm?

Could I not assume that the door does not deflect and all the load is transfered to the hinges? This would be the worst case sceneario, but if i can live with it then ok.

 

[ccw]

Hi ddeck,

Isn't the robot going to pick up the heaviest part it can swing to hit the door when it tries to escape?

 

[ddeck]

Yah, I'm using the max weight the robot can carry.

The thing is the arm weight around 60 lbs, but its not concentrated at the point of impact, and therefore i don't think i should use all of that mass for calculating the Force generated.

And the other thing is that what is a reasonable time which the collision occurs. Opposed to my original thinking that i could neglect the deflection, i think i need to determine the deflection of the door, which with the speed the robot is travelling at i could get the impact time.

 

[mrMikee]

I would use the work-energy equations. The kinetic energy is a function of the moving mass and velocity, and the work done is a function of force and displacement. However, the force isn't constant but varies with displacement like a spring. The unknowns are where on the door impact occurs and whether the door remains elastic.

Regards,
-Mike

 

[MintJulep]

Why don't you just tell the robot not to hit the door?

 

[ddeck]

Well of course we are going to tell the robot not to hit the door, but doesn't mean that something can't go wrong and it hits the door.

And the door happens to be partially transparent and people will be observing the cell in action.

Therefore, we need to be able contain the robot within the cell.

 

[mrMikee]

Here's a solution for a spring-mass system which probably can be found in most dynamics texts.

work = change in kinetic energy ''general equation''

where:
work = 0.5*k*x^2 work done by spring
k.e. = 0.5*m*v^2 kinetic energy

and:
m=mass
v=velocity of mass
k=spring constant
x=deflection of spring

Solve for x and then calculate F=k*x. You will need to determine the value for k to do this, probably something from Roark and Young.

This approach is of course a simplification of relatively complex problem. And there are other variables that might be important. I would assume for example, that in addition to the impact of the robot into the door you should consider the possibility that the arm is still being driven, in which case that force would need to be added to the impact force. And so on.

-Mike

 

[MintJulep]

I suspect that you will not tell the robot to hit the door. This is different from telling the robot to not hit the door.

For a back-up, I would suggest a light curtain on the robot's side of the door that removes power if the robot gets too close, with the secondary benefit of stopping the robot if someone opens the door and walks in.

 

[ddeck]

That is an option, but I don't know if a light curtain will work in our situtaion. Space is a constraint and would could only put a curtain under 1" offset of the door.

And another point that my boss brought up, if your robot is flying at its maximum speed, and hits the light curtain, power is cut, but the light curtain doesn't take away the momentum the robot currently has and therefore would need to be told to brake or else it goes until it hits something.

I don't know much about how a light curtain works, do they remove all power to the arm, or do they tell it to stop (turn on the brakes)? If the light curtain tells it to turn on the brakes becasue it isn't listening to its programing, what makes it listen to the light curtain?

 

[tygerdawg]

The best approach is not to construct a cage to keep the robot IN the robot workcell, but to keep idiots OUT of the robot envelope. The motion of the arm can easily be controlled through existing joint hard stops, sensors, etc., that are much easier to deal with than vague calculations.

Upon impact, the robot joints are forced to absorb a reaction torque. Each joint more or less depending upon configuration, speed, payload. The servo drivers constantly compare each joint encoder's TARGET position versus ACTUAL position. If this value is larger than a predetermined limit, the servos are killed to save the equipment (as much as possible, anyway). If you calculated the work required by a barrier to absorb an impact, the reality is that the area under the curve would get truncated somewhere along the way. You would have over-designed your barrier.

I have designed machinery with typical machine guards constructed of aluminum extrusion, steel fab, polycarbonate panels, or woven wire panels. If your robot is so big and powerful that it can get past any of those and cause injuries to humans, then the workcell is a legal liability and needs to be redesigned. Operators need to be denied entry to the work envelope except under controlled conditions.

ANSI/RIA 15.06-1999 robot safety standard is a good reference for designing robot workcells.

TygerDawg

 

[MintJulep]

A light curtain, or any other limit switch simply provides a signal.

How that signal is used is up to you. It can be used to turn off a relay to remove power to the robot. It can be used as an input to the robot's controller with specific code to execute if that input changes. Or some combnation of the foregoing

The brakes should be energize to release, so if you turn off the power, the brakes should come on.

 

[gwolf]

Here's a simple method:

Find out by experiment how fast the robot can hurl the heaviest AND the lightest weights at the door ( Energy = 1/2mv^2)

Make a simple mesh cage to fit over the door but omit the glass at this stage. If the door opens inwards you can react the cage impact loads on the door frame rather than the door.

Drop things on a test cage/door/hinge assembly with masses and speeds (speed equivalent to height dropped) equal to what you measured off the robot.

The cage will bend, if it bends enough to have touched the glass you will need to beef the cage up. This all assumes that that the door frame on the working enclosure is adequately fixed to the machine bed.

I like this idea more than doing an FE model, it will be fun, realistic and cheap. Furthermore you could ask the robot if it would like to participate in the trial rather than dropping weights onto the cage.

Just thought, if the objects carried are pointy then you need to make the cage mesh smaller than the points or stand the mesh well clear of the glass.

I wish I was doing this, very interesting.