Telepresence Robot (XMOS Challenge)

Let's Make Robots Robots
1

This is a Telepresence robot, it is intended to be connected to the Intenet (Wifi) and allow a remote person to interact with the (remote) environment. Whilst not in telepresence mode it will interact with people locally.

The project is based on the work of the leaf.org with some components from NASA's Claraty project.

High level functions, planning/mapping/video processing is done on a mini-ITX with cut down Linux (puppy) running software ported from the leaf project and Calarity.

Low level functions such as Physical interface management, gyro, servo, motor control and low level sensor fusion and filtering via XMOS processor(s) communitating to the mini-ITX via serial.

Power via two battery packs 1 x 12V NIMH for the logic and 2 x 6v NIMH for drive.

Drive is via 4 320 series motors using spur reduction boxs,

Steering using all four wheel, the principle challenge is that it is intended that this robot to be able to navigate stairs by alternatively balancing on the front and rear wheel sets, the geometery of the "leg"s may be altered to increase and decrease the height and move the CG over the wheels.

TPR_XMOS_Front.jpg375×500

The above photo is the back bone of the robot, I am using the head off a Robot Quad, I contacted Wowee and they sold me the neck and head, I could not bring my self to hacking mine! The head will contain 1 x Gameboy Camera and one Wii remot IR Camera and the mic. The head will be mounted to allow yaw, pitch and roll for expression. The PC is running puppy linux with the the graphics removed (all the apps at this stage) and boots from flash. I have built a PC running a similar build except it boots from disk and has the GUI interface left in.

TPR_XMOS_Rear.jpg296×500

This is a view of the rear, the mini-ITX Board is mounted to a fiberglass sheet which is attached to the backbone using fokker clips, top hat shaped flanges are turned for the top and bottom mounting.

This is one of the wheels, they are Roller Blade wheels with and Alu hub fitted, RC car roller bearing with an RC car gear as the final drive, the wheel mount has a gearbox (18:1) to give approx 49:1. Motor control will be softstart & constant G. I think the Plastic gear will be ok, I am a little concerned about power especially when balancing the motors will bang back and forth a lot.

Act as a virtual person remotely, allowing a person on the Web to interact with others in an unstructured environment

  • Control method: autonomous (very), Wifi connection to Internet
  • CPU: XMOS processor(s) performing the low level motor control, pose control etc, mini-ITX PC running linux performing the higher level functions
  • Operating system: Pupply Linux
  • Power source: 4.5AH 12V NIMH for logic, 2 x 6V 2.5AH NIMH for drive
  • Programming language: C++, C, XC, Lisp
  • Sensors / input devices: Gameboy Camera, WII Remote IR Camera
  • Target environment: indoors mainly
This is a companion discussion topic for the original entry at https://community.robotshop.com/robots/show/telepresence-robot-xmos-challenge
2

I don’t see much of a

I don’t see much of a problem with this, a design and details is a good idea! Gets people excited and interested for more! I’m not really sure if it’s “too early” to put up as a robot, but it is a “Work in Progress”.

But just posting a robot, that’s completely empty, with no details or pics, etc etc… agreed, delete!

But then again, I’m not an admin either, so up to them!

 

But regardless, I think it looks good! I’m interested to see how well it turns out! You could also do some temperature and light logging, just for the sake of knowing. :stuck_out_tongue:

3

More than Sketch Up

thanks

I will put somephotos up when I get home, I did not have them with me at work, I have the prototype processor set up and I am turning the planetary gearboxes at the moment.

But I will be more careful.

An interesting question is when is a robot a robot and not a collection of parts? I suppose it is important to have a stanard

of what appears where.

all the best

Cliff

4

A question - have you looked

A question - have you looked at using a netbook style system rather than an ITX board and seperate display?

 

You get everything like CPU, webcam, display, and a keyboard for those times talking is an issue , all integrated, just add an arduino and some sensors/motor controllers via USB-serial. You could probably butcher an old office chair too…

5

Note book vs ITX

Hello Graphworlok,

Yes that was the intent in original design, as can be seen in the diagram. In using one of the netbooks in may respects the choice of a netbook simplifies a number of things and complicates others, with the mini-itx I am using I have 2Gb of memory, and more clock cycles than I will need for the first 12 months anyway; this gives me a little more head room and performance.<o:p></o:p>I am running Puppy Linux at the moment I am cutting the bits out of the Kernel I don’t need. I have started porting some useful modules from XP to Linux. <o:p></o:p>A word of warning to anyone considering using an Mini-ITX, a lot of the DC power supplies designed to allow the use of the computer in the car do not regulat the 12V, some Mini-ITX’s have onboard regulation on the 12 volt rail so this is not a problem; a lot of others don’t and you will have problems and could cook a board. A case of knowing what you are connecting together!!<o:p></o:p>

Regarding the later post talking about the use of the XMOS processors, this project uses multiple processors, low level IO processor which service such things as motor control, servo management; pose management leaving the PC to do the higher level functions, a little like the Mendulla and Cerebrum.

The first pass design used Arduino for the low level, Arduinos are great but I decided that I wanted to use comutator noise on the drive motors instead of encoders and stall current detection for bump detection. I moved to considering the DSPIC platform, these are great chips and work well, but controlling 16 motors would require a sea of them; then I saw the XMOS processors, the multi-thread event based processing solves a lot of sticky problems. <o:p></o:p>

I need to sign off now and take some photos of the robot bits so I can post them.

<o:p>all the best </o:p>

 

this project was not originally intended for the XMOS processors, initially I looked at PIC’s

6

The thing is, it’s for the

The thing is, it’s for the XMOS challenge so he has to use the XMOS processor :wink:

 

(Edit : sorry, i think i misunderstood few things, nevermind)

7

320 series?

What are the 320 series motors? Tell more.

8

320 series

Hi Calculon,

a long time ago, when RC cars started the motors where named based on the outside diameter of the motor (dia x 10 in mm).

A typical RC car these days uses 540’s ie. 54mm dia.

I may go to 540’s, I want to be able to balance on either pair of wheels, so I need to over power a little to get the response rate up for the balance.

 

all the best

/CW

9

Old office chair

Hi Graphworlok,

A little more,

the software I am using (see www.leaf.org) has voice recognition, and the robot will be connected via Wifi.

The H shaped chassis is to allow the robot to change it’s hight and allow it to balance on either pair of wheels.

It will (well I hope) be able got up stairs by balancing on the rear wheels articutlating the legs and then balancing on the fronts (next stair up), this is an experiment, it works on paper! we will see if it works in practice.

all the best

Cliff

10

I assume that the mobo will
I assume that the mobo will eventually be in a waterproof box? Looking good, what are you going to do about shock/suspension though?

11

Water Proof

Hi Krumlink,

I have not thought about the box yet, its for indoors at the moment.

Suspension is via (at this stage) rubber dampers in the hip joints.

When balancing the suspension works againts me, flexible inverted pendulms are a can of worms!

thank you for the questions

all the best

Cliff

12

interesting

Good luck with porting the leaf software over to xmos… i was attempting this about 5 months ago with arduino.

It seems when you ask questions on their group boards about porting the leaf software over to arduino, they delete your messages. I have posted about 20 messages on there, of which 5 were actually posted, and they did not relate to arduino. The rest of the posts did.

I had considered using leaf in combo with xmos myself for my xmos challenge bot, but I just dont think I am capable of reporting the nav and control software from the leaf project.

Now if you wanted to go in together to see if we can port it to xmos , I would be more than happy to help, as I have already learned quite abit about the leaf project. I really think that the porting of all the work the have already done to a more common processor, or even one that is more avaiilable is a good thing. The processor they are using is $$ and not many places carry it.

Good luck with your adventure in to leaf land.

As a side note, The leaf project is pretty cool, you should all check it out. I have the leaf software running on an old pc now, and it actually does facial recognition, voice recog, and a lot more.

very cool.

LOL, i recognize the head in those pics too! I have it on three of my bots. Think there are a couple of pics of mine on my robots page.

 

13

Porting Leaf

Hello Grinan,

If you want I am happy to share the ported code with you.

I will be porting the controller code, but mine will be significantly different to Leaf as I am controlling 4 wheels (drive and steering) leg positions and spine as well as the head pose, very different to leaf, the robot has two distinct kinematic modes, 4 wheel mode, static balance and a 2 wheel inverted pendulm mode.

The south bound side of the interface will be different, I will be using the FTDI interface in the direct connection mode.

all the best

cliff

14

Wowwee
What kind of hardware is in Wowwee’s head? Cost? May be useful to me as well…

15

Would love to see your

Would love to see your port. I really wanted it ported to an arduino mega myself… that’s my processor of choice… but there was just sooooo much code in the nav and command files that i was very intimidated.

The Quad head from wowwee that I had actually came off a quad that met an untimely end… it had an ir detector/receiver, a few leds and an ir led. That’s all I remember being in there. On a side note, the paralax ping units fit the head perfectly, with no modification. Just pop them in with a little hot glue, and it works perfectly. I never thought about buying one right from wowwee… that makes a lot more sense than stripping one down.

 

16

Ambitious project!

Ambitious project! I will look forward to see more of your progress… we can learn a lot from here!!!

Keep it up :wink:

17

Wowwee’s Head

Hey Benbo,

Sorry to take so long, work is trying to kill me.

I must say the guys at Wowwee were great, from memory it cost my about $40 (aud) including freight.

You get the head and neck, there is a gearmotor/servo in the upper neck, and in the hear there are a few LEDs and IR detectors with a Circuit board for them.

all the best

CW

18

Ambitious

Hi Guibot,

thank you, I hope I will produce some bits and peices worthy of note.

It is ambitious, I have been building robots for awhile but they have been reasonably simple and stuff which have been done before.

I was planning on building this robot for more than twelve months, then I came across a YDM video and thought I gotta have one of those so I got side tracked a little. The YDM video lead me to LMR so alls good, it is a great site.

thanks again

CW