DaVinci Project results

Insertion Task Results

Simple Rotation

To explore the sensitivity of the peg motion during a simple push, many trials with random support point locations were simulated and compared to experimental data:

Exp. #	# of Simulated Trials	XY Mean Squared Error	Max XY Squared Error	Theta MSE	Max Theta SE	Data
1	121	2.91129870924292	14.9980285216273	0.0694373764268487	0.233377566947717	pdf, txt
2	150	2.44577102713617	6.07511766473507	0.0691177627689829	0.247736664621041	pdf, txt
3	150	5.43774761445843	12.1004513252581	0.0607214011321805	0.176349740743197	pdf, txt
4	115	2.93100674007574	10.4956492662992	0.0723911137399792	0.322731671574136	pdf, txt
5	123	3.1907914989465	9.27052596545321	0.0569100313737789	0.248886714417452	pdf, txt
6	111	7.72839390190476	13.9739943659346	0.0787997184310462	0.241871530368466	pdf, txt

Note: Units of this data are 0.1mm for the XY errors and radians for the Theta errors

I'm showing the mean and worst error over all of the simulated trials there are some support configurations that produce motion very close to experimental results. I am going to pick one of these support distributions and use it as the right one from now on. In retrospect, perhaps we should have varied the location on the peg where we pushed from experiment to experiment to make things a little more interesting.

Insertion Plan

An insertion plan that worked in simulation and hardware: visual of experimental insertion, visual of simulator insertion, X,Y comparison of the paths (similar), X,Y,Theta vs Time comparison of experimental and simulated.
[OLD] X,Y,Theta Comparison of

experimental

vs. simulated insertion.

Feeder Task Results

Really simple friction model

I implemented the really simple friction model we talked about for the feeder example (due to the fact it is sitting at a 65 degree angle out of our simulated plane). To find frictional force:

Get opposite direction of body's velocity => frictionDirection
Calculate friction force = mu*m*g*cos(65deg) => frictionForce
Pext = Pext + frictionForce*frictionDirection

Right now I'm only applying this frictional force when the body is in free-fall (ie not in collision with the feeder structure). The raw data from this simulation is here.