Bing 3

[Robo1]

Looking good.

thanks for the link about the pressure sensors. how well are they working can you use them to see if his foots flat on the ground, did you have to put rubber over the top.

bren

[DerekZahn]

As for how well they are working, they seem to behave the way I would like but since Bing is still only half finished I don't know for sure how well they will work in practice.

The sensors are completely flat and rely on pressure in a central zone to get readings, so I would think that some sort of extra addition is necessary, like attaching a 1 cm diameter rubber "foot". I have some concerns in this regard -- first, what is a good attachment mechanism? My first thought was to just glue some rubber onto the front, but since the sensors are plastic that has some limits on the glue types (cyanoacrylate "super glue" would probably eat right through the sensor). Also, if the glue is too stiff I wonder if that would alter the response of the sensor.

Then there's the issue of shear force. I wonder if over time shear forces caused by scuffling feet could cause the sensor to delaminate a bit and alter its response.

To help deal with this I decided to try to attach 1 cm diameter, 3mm thick rubber pads to each sensor, then attach those pads on the other side with a large single plate. Earlier in this thread I posted a link to a picture illustrating the construction of this, and the very first photo in this thread is a picture of the plates. I have not attached them yet. I am hoping that this type of construction will distribute the sideways forces to protect the sensors a bit. But it's all theory at this point.

If all four sensors are showing a pressure reading, the foot has to be basically flat (assuming the ground is flat) or else all four corners would not touch. Further, differences in pressure readings should determine whether the weight of the robot is equally distributed across the foot, which I think will be helpful in maintaining balance. Measuring the differences in total readings between one foot and the other will also give information about the robot's balance. Watching for changes in the proportion of readings from different sensors should be helpful in determining whether the robot is wobbling or falling. And of course simply watching the sensor values should determine whether the foot is on the ground or not, which should be helpful in walking.

At this point, that's all theory though, I have done nothing to prove that this will work and I don't even have any well-formed ideas about how to do this. I'm just building something that I can spend most of 2007 trying to program to do these things.

I think it will be fun, but I am not certain I will succeed.

[DerekZahn]

A long time ago, shortly after the HSR-5995TG servo started to become available but supply was a problem, I got a good deal on a bunch of JR servos so I grabbed them. Since I have them I have decided to use them in Bing's arms.

The biggest drawback of these servos is that although their mechanical range is not horrible (about 180 degrees) the range they actually achieve with applied pulses is less than 120 degrees. That's not sufficient for the arms. So time for some customization!

Opening up the servo reveals the guts. Unlike the Hitec servos, the motor is not soldered directly to the circuitry, which makes this servo much easier to work with.



bigger: http://happyrobots.com/jr1.jpg

Prying up the electronics board shows the connection to the potentiometer, and a cavity above the potentiometer with enough room to add some stuff!



bigger: http://happyrobots.com/jr2.jpg

To increase the range, I want to change the resistance of the potentiometer so that a larger range of motion corresponds to the same voltage values on the central tap of the pot. This can be adequately done by adding two resistors.

Another thing that has to be done is isolate the potentiometer from my external A/D circuitry. Although an A/D port is pretty high impedance, I have found that hooking the central pot voltage directly to an external microcontroller can change the circuitry and this the response of the servo. To get around this, I isolate the value using an op-amp configured as a "voltage follower", which is just about the simplest thing you can do with an op-amp and provides a super high impedance isolation. So here's the circuit that I want to stick into the servo:



I have a pile of op amps and resistors laying around, and some blank circuit board material, so I decided to try using the mill to make the circuit boards so I didn't have to order anything and wait for delivery. Here's the mill working on it. It is using a 0.012 inch end mill (0.3mm).



bigger: http://happyrobots.com/jr4.jpg

And here's the resulting board. The three holes at the top will attach directly to the potentiometer pins, and the 5 holes at the bottom are for the three potentiometer wires, one wire to go out to my external microcontroller, and one +V wire for the op amp power supply.



bigger: http://happyrobots.com/jr5.jpg

To make life easier, I tin the board with some solder, as soldering components directly to copper board is not always as simple as I'd like.



bigger: http://happyrobots.com/jr6.jpg

Surface mount resistors would work better, but I had some little normal ones sitting around so I used those.



bigger: http://happyrobots.com/jr7.jpg

Attaching and tinning the wires gets everything ready for the implant:



bigger: http://happyrobots.com/jr8.jpg

And here is the result, with the surgery complete. I now have the position feedback on a wire, and 180 degrees of travel in the servo



bigger: http://happyrobots.com/jr9.jpg

One servo done, 7 to go. Sigh.

[DerekZahn]

One more bit of progress... The design for Bing's upper torso ended up consisting of five different plastic pieces (somehow I never manage to make things as simple as I would like). The first of those pieces is finished:



bigger: http://happyrobots.com/bing3m2.jpg

[Robo1]

Looking sick

will have some new components built buy the end of the week. Made a mistake ordering new componets order 3xCM2 and 1xDX-117 instead of 3xDX-117 1xCm2. will put me back a couple of weeks but mistakes happen

On your HSR-5995TG did you have trouble with the pot conversion. I ask as I'm thinking about doing this mod on my kondo servos and wonder did you have trouble with the voltage also wondering how it affected the performace of the servo e.g. changed any charactistics.

Bren

[DerekZahn]

I haven't noticed any effect from tapping the pots. But I do use the same op-amp voltage follower to isolate the signal on the Hitec servos also -- I just don't put the circuit inside the servo. Instead it's on the little daughterboard for the microcontroller that I showed earlier in this thread.

Good luck with your construction, take pictures!

[DerekZahn]

All the JR servos are modified now and some other parts have been finished as well. Here's the various parts that make up the upper torso:



bigger: http://happyrobots.com/bing3n2.jpg

I think I'm 2-3 weeks away from having Bing finished mechanically and electrically, then I can get to the hard stuff!

[DerekZahn]

While I keep working on little parts for Bing, I also got some work in on the UI program. Here's its current state:



bigger: http://happyrobots.com/bing3o2.png

Sorry for the size of the picture, I have a large computer monitor and I'm using all the screen space I can.

The numbers in the graphs don't mean anything... the program is not plugged into Bing. I'm just testing the graph drawing code by feeding it random values for now.

[DerekZahn]

I haven't been posting as many updates but that doesn't mean I've been idle... it's just that the construction methods are just about the same as stuff I've shown before. But Bing's arms are finished:



bigger: http://happyrobots.com/bing3p2.jpg

My goals for Bing 3 (besides just having fun in general) are to compete at RoboGames and have at least a chance at qualifying for the Robo One #12 event. To do those things, Bing has to be able to:

1) Walk in a couple of different ways
2) Stand up after falling
3) Make a punch of some kind

2 and 3 are helped tremendously by having arms that are at least a little bit useful, but I'm most interested and focused on walking, so these arms are pretty simple and boring. Someday in the future I'd like to add a rotation axis and a gripping hand of some kind, but I can't focus on everything at once so I'm going with simplicity.

Now I go back to electronics for a little while -- I need to get the processors for the arms set up and working, and I am going to go back and add 10 current sensors to various parts of Bing, which means some surgery.

[DerekZahn]

And while I'm playing with the camera, here's a picture with the parts put in place (the top half of the torso is not actually attached to the bottom yet, it's just placed on top for the picture)



bigger: http://happyrobots.com/bing3q2.jpg

I think it's getting pretty close now!

[DerekZahn]

One last shot: headless fake shot of Bing standing:



bigger: http://happyrobots.com/bing3r2.jpg

[JonHylands]

Looks very sweet, Derek.

- Jon

[Robo1]

What sort of head are you going for.

One just for looks or is it going to have a camera or something in it.

looking go when do you think it will be walking by.

I've just finished building some parts for my bot just need to spray them and try them out. DX-117 are out of stock so have to wait for some more to come in before I can put the legs together.

bren

[DerekZahn]

The head will just be for looks, at least for now. A camera or sonar unit would be fun but if I try to do too many things I'll never actually achieve any of them, so I'll save that kind of enhancement for the future.

I don't know how long it will take to get it walking, but I know it's going to take a long time because I want to think about it carefully while working on it. RoboGames is in June so "hopefully by June".

[RobotJay]

Derek,

Fantastic work. Nice build log, too. I am also going to compete in Robo-1 at the RoboGames this summer. I think there will be some good competition there between you, Pirkus (robot labs), and Rook's Pawn (Matt Bauer of Bauer Independents). I hope anyone following these boards can come to San Francisco this summer and watch. Regardless of how your bot functions, it looks TASTY. Mmmm... Keep up the good work and keep us updated with the pictures.

-Jay