View Full Version : Sir Kitt robotic TV presenter

05-07-2009, 11:51 AM

Apologies for the cheesy video I made, makes me laugh every time I see it.

YouTube - Sir Kitt : I've not come in peace, i've come to dance

Well, where do I start?

Around November 2008 a TV production company came looking for an enthusiast who could build a robot to interact with guests backstage at a large music awards ceremony. I put myself forward for this as I thought it would be a good opportunity to motivate me into producing something decent for a change :)

The TV producer was looking for a humanoid that could wander backstage and interview guests with a camera in its head sending back a feed that could be recorded for the show. Unfortunately, the backstage area was enormous with uneven carpets, cables, etc so a humanoid on its own would not have gone more than a metre before falling over.




There was obviously a need for a vehicle that could move the humanoid from guest to guest. Luckily I had an unbuilt Lynxmotion 4WD1 rover kit and a partly finished Manoi that I had never managed to find the time to finish. So after a long brainstorming session I set about trying to build a robot system as below.

Robotic armchair

Fully adjustable speed and direction
Lightweight body
System for locking and unlocking the humanoid from the seat to stop it falling off
Lightweight LIPO batteries
Video camera to provide second feed
DV Camcorder to record video and audio feeds
Remote speech playback unit
Amplifier and speaker system with mesh grille
RC receiver
fused busbar power system
Hazard flashers
External kill-switch


Finish assembly of Manoi and paint up the shell suitably
Primary Camera in robots head
Upgrade batteries to performance LIPOs
LED eyes and chest lights
Angle head to look upwards
Add gyros, balance and create specific moves


Belt mounted AV receiver, LIPO battery and 2nd DV camcorder for Manoi head cam and audio feed recording.

Remote controls for chair, Manoi, speech unit and camcorders

Charging and tape swap systems

Mountain of spares, batteries, etc and a system for controlling them.

Lynxmotion Robotics (http://www.lynxmotion.com)
Manoi AT-01Humanoid (http://translate.google.co.uk/translate?u=http://www.kyosho.com/jpn/products/rob%20ot/at01/at01.html&sl=ja&tl=en&hl=en&ie=UTF-8)

Oh, and did I forget to mention, I had only one month from the final ok from the producer to get the robot built and tested, all for no money as I was to be building this for the 'experience' :(

First thing was to check out the dynamics of fitting together the Manoi and the rover which would dictate the shape of the armchair. The main body of the chair was made from thin MDF board, pinned and glued together using thin wooden battens where necessary. I took my time and made sure that none of the fixings could be seen, I wanted the outer skin to be as blemish free as possible.



The rover was assembled easily and I wired up the motors with cable from an old PC PSU. I tested each motor in both directions so I could create two matched pairs, one pair for left side and one pair for right side. If the rover pulled to one side, it could easily be trimmed out.


The chair was then mated to the rover and there started many hours worth of staring into the back trying to envision where everything had to go. Once happy, I set about adding mounting points and restraints to the MDF where I would need to secure the components to the shell.



A Sabertooth 2x10 speed controller was installed into the rover and quick release wiring allowed the rover to be separated from the shell where the RC receiver and power distribution were located. This comprised of an old USB extension cable cut in half and used to disconnect the RC signals and a Tamiya battery connector for the power.

Dimension Engineering Sabertooth speed controller (http://www.dimensionengineering.com/Sabertooth2X10.htm)

I created a safe space for a LIPO battery pack in the heart of the rover and made sure that there were no nails, screws, etc that could move and spike the pack. Tamiya connectors were used again to connect it to the main power switch/fuse.


After some fun testing and chasing the cats, I moved on to the shell. The first idea I had was to drape a Union flag over the whole thing and tailor it to fit, but try as I might, I couldn't get the angles to work correctly so I resorted to a complex paint scheme involving masses of tape and newspaper and can after can of spray paint. In the end it came out much better than I had thought it would. There was no way of getting the proportions of the flag to look 100%, but it was as good as I could get it.






Next up was the electronic speech module. I decided quickly that I couldn't fit a suitable system and high volume speaker within the Humanoid, so all of it would be contained within the rover. My idea would be that the armchair would act as a second camera man, dropping off the Manoi and backing off a little to get the star and robot in range of its second camera. As it wouldn't be far away, it wouldn't matter if the robots speech came from it as well. Also, the line output of the speech unit could be passed straight to the internal camcorder for better quality recording.

I went through literally hundreds of ideas for the speech unit, ranging from pda's and AV links to actually changing my voice and transmitting it. After about a week of pouring through circuit diagrams and module specifications, I decided on using a remote keyboard to send data to an MP3 playback module in the rover. The only trouble was that nothing actually existed that fit the bill so I had to mix and match rather ingeniously to get it all to work.

The heart of the system is a Quadravox MP3 playback module, the QV606 which can handle hundreds of small MP3 files and benefits from a serial input system which had a very simple protocol that allowed me to connect it to an 'Easy radio' serial transmission system. Luckily, everything worked together and I ended up with a small box that housed the Aerial, Easy Radio receiver, QV606, pre-amp, power-amp, volume control and all the connectors. The only leads needed were the power in and the speaker out !!

Quadravox speech module (http://www.quadravox.com/QV606/QV606.htm)
LPRS Easy Radio Module (http://www.lprs.co.uk/index.php?cPath=21)




The amplifiers were cannibalised from my daughters' old pink CD boombox. She was devastated when she found me disembowelling it to get to the power amp, but I said that although the CD player was dead, it would live on ...

The 'Easy radio' serial link was very easy to set up and worked flawlessly. I tried a couple of other RF modules before and had limited success, mostly due to baud rates and packet encoding. The addition of a short stubby aerial at both ends meant I had an amazing range of at least 100 feet for such a low powered system.

Next stage was the keyboard transmitter unit. I searched for a method of having a keyboard or keypad unit that would output RS232 at the correct baud rate for the Easy Radio transmitter module but ended up forcing together things that had not been tested before. I decided to use a PC keyboard as it had more than enough keys to ensure I could get to all the robot words and phrases as easily as possible. After lots of more expensive AT - RS232 converters, I lucked upon a Radio amateur who developed a small 8 pin IC that converts PC PS2 signals into RS232 at 9600 baud.

PS2 keyboard to RS232 converter IC (http://k1el.tripod.com/ATKBD.html)

This needed no external components and created a very simple solution that would take up little space and would also require little power.

Looking at the size of all my components, I could see that I may be able to get them all inside one of the new micro PC keyboards that are around. I needed to find one that had proper PS2 support as well as the usual USB support and after a couple of wrong-uns, ended up with the one in the photo.

Taking the back off it, there was just enough space to make this very neat indeed. I added the converter chip, the transmitter module, 5v regulator and copious amounts of hot glue to eventually make a really neat looking unit. The aerial poked out of the top and there was a PP3 battery connector to the rear for a 9V battery (the unit worked for the whole 8 hours on a single battery).


Being a really neat smart little unit, I thought it would be great if I could integrate it with the PS2 controller used for the humanoid, so I found a way to use the long screw holes in the back of the PS2 controller to friction/gravity fit to a couple of arms which would support the micro keyboard at a comfortable angle.

The unit was now a self contained humanoid and speech controller. I could move the humanoid around and stretch out a finger to make him say whatever I wanted from a pre-programmed list of up to 255 phrases (the shift and CTRL keys acted as modifiers). That system alone was probably my best ever achievement in electronics and I am immensely proud of it. This is of course a summary of what was involved, just figuring our protocols, circuit diagrams, programming of the speech unit and various compatibility issues took for ever and would probably justify a build report of its own if there is enough interest.



I used the Microsoft Sam speech engine, modified by using Alien Speech to produce the robotic sounding voice and got busy encoding to MP3 anything I thought would be good for conversation or funny. The names of all the stars were encoded so I could string things together, i.e. Hello - Bono - are you nervous, etc. All this was then uploaded to the QV606 module and assigned to keys on the remote. I printed out sheets of info in a tiny font to stick to the keys on the keyboard to remind me of what was where. I also cut out cardboard copies of the PS2 Humanoid controller which are held just above the remotes to guide me through all the available moves if I got lost.

The receiver/amplifier module fitted nicely on a shelf in the shell and I connected up the power and speaker connectors. The speaker itself is mounted in the top of the shell behind a grille from an old desktop PC speaker I cannibalised. It's all painted over so it fits in with the pattern nicely. I made sure it was away from other cables and the DV camcorder to avoid interference issues.


I bought a night/day colour board camera which poked out of a hole through the front of the chair, this was held in with a bendable bracket so I could adjust its angle. The output was fed, along with the speech line output, to the DV camcorder. A power cable went back to the rovers' main power rail.

I searched for a small, cheap, powerful flasher warning system for the robot which was insisted upon for H&S reasons, but as with everything else, I decided to make my own. I bough 6 ultra bright amber LED's which I need to point front, back and outwards, one set for each side of the robot. I then went to my local DIY shop and searched everywhere for a small amber bit of plastic to mount them in. I found some small battery powered handheld fans for 99p each which had a see-through amber plastic dome on the end which I thought would do nicely. I made a quick 555 timer circuit to flash at the correct interval and powered it all off the main power bar.


The main power switch was from an old printer and I combined it with a 10 amp in-line auto fuse to stop things getting smokey. The external kill-switch made from plastic from an old kids toy which poked through the shell and was secured with a spring and washer. All it did was to poke through the shell to the main power switch inside.


The Humanoid is held onto the chair using an upright metal post that he lowers himself onto, there is a similarly shaped hole attached to his rear (oh er) which ensures it secures him well. There are also two rubber posts for him to rest his feet on to stop them dragging on the ground (made from two plumbing washers).



The humanoid was completed and I decided on the colour scheme. Painting was easy enough with special RC polycarbonate paints but I had some leakage around the edges of the masking tape I used which needed to be touched up. Two gyros were set up to keep him stable and a flashing multicoloured led was added to his chest to shine through the shell.




His head needed complete modification as I needed to mount a wireless camera inside and a power converter to provide it with the correct voltage. Unfortunately, during testing it became apparent that movement of the servos was causing interference in the video signal. I ended up mounting a PP3 battery in the head which meant an on-off switch and holder was required. I also had to make a wedge shaped shim to angle the head upwards towards the guests. Two ultra bright blue LED's were added for his eyes. The head shell was modified to ensure I could quickly access the battery by removing a couple of clips.



This all added weight to the head and meant that none of the standard moves would work. I had one night to completely re-code the bare essential moves that I would require on the next day.

Another big outlay for this project were the batteries. I had the following

6 LIPOs for the Manoi
4 LIPOs for the armchair + chargers
6 LIPOs for the belt mounted AV receiver
20 PP3 batteries for keypad remote and Humanoid head camera
20 AAs for the PS2 remote, camcorder remotes and the wireless microphone
10 DV camcorder batteries. + chargers


The tapes I used were bought from Ebay in bulk second-hand and proved to be an excellent buy. It was quite amusing to see that they must have come from a wedding photographer as none of them had been wiped. Many hours of footage of customers' special days had to be wiped off !!

All of this was worked into a month which meant every spare moment was spent working on the robot, all weekends and every week night up till 2am. My poor wife had to suffer the complete take-over of the dining room, which was turned into a workshop for the month.


Eventually, it all fell into place and all the bugs were ironed out on the day before the filming.




On the day, I had to make up a system for the rotation of batteries, etc which was governed by the DV tapes. They lasted 60 minutes which meant that at 55 minutes I had to do a pit stop and swap over all the tapes and batteries and get the charging cycles going again.

The event itself was amazing. I arrived early and was shown around and given a small area to set myself up. I spent all day roaming the corridors and dressing rooms interviewing some of the biggest stars in the world before and after they did their sets. I quickly settled into a routine but had to master most of the controls on the fly.

After 10 hours of filming, it was all over. I met some great people, both in front and behind the cameras and walked away with amazing experiences and memories.

Unfortunately, to my horror, the TV company ended up replacing my whole segment from the show with normal interviews so none of the footage was ever broadcast. To say I was gutted was a vast understatement. I pulled myself together and harangued the media company for a copy of the tapes which they eventually agreed to, but I had to sign a VERY scary contract beforehand restricting the footage to my own personal viewing only.

All in all, I still had a great time and the video interviews are amazing.

Hopefully, now that I have all the equipment built and ready to go, I can get some other work for my robot TV presenter and continue to enjoy the fruits of my labour.


05-07-2009, 12:04 PM
Great writeup! Thank you much for sharing, pretty neat idea for socially interactive bots :)

Paint job on your Manoi looks fantastic as well.

It's a shame that after all the hard work you put into it, it got cut. Such is showbiz though eh?

+Rep from me.

05-07-2009, 02:46 PM
Ok I'm gonna read through this post now, I've only gone over the pictures yet, but I'm willing to say : damn...

That's one helluva job! :D

05-08-2009, 03:22 AM
What a creative solution! The vehicle transporting the humanoid is the takeaway for me here. Nicely hibred to solve the problem!

05-08-2009, 03:59 AM
I have seen this robot and transporter actually working and it's amazing!

Brilliant work Orac!


www.conceptioneering.co.uk (http://www.conceptioneering.co.uk)

05-08-2009, 08:59 AM
I love the concept of the motorized chair. Very clever solution!

05-09-2009, 03:22 PM
Thanks, I'm getting together a more sensible video this weekend to show the kind of things you can do with this robot. With 255 phrases to store and pick from, you can really get a natural sounding conversation going.

05-13-2009, 06:33 AM
I have a bit of a problem with the electronics.

For recording at the Brit awards, most of the attention was on the stars being recorded by the onboard cameras. Going forward, I also want to record video of the robot itself.

To increase the sense of interaction, I felt I needed to add an LED mouth to the Humanoid to give more of an impression that the voice was coming from his mouth and also to provide something visual instead of having him staring blankly into the camera while speaking.

I need to take the audio output from the chairs MP3 player, set a threshhold and then pass the pulsing output through a transmitter to a reciever in the Humanoid which will then turn the pulses into LED output at the mouth.

I have a cheap AM tranmitter and reciever unit set up, the output is connected via a transistor to LED's and all works well when I apply test pulses (via a 555 timer) to the transmitter.

The MP3 module has an L and R line output. I have fed the L output to the pre-amp and then via another amp to the main speaker.

However, when I connect the transmitter input to the R line and audio ground output of my MP3 module in the chair, I get big sound distortions in the pre-amp (which is fed from the L line out). It seems that the cable between the MP3 and the transmitter is picking up the AM broadcast and feeding it back somehow to the Pre-amp even though it is normal phono coax and grounded at both ends. The Transmitter is powered by a 9v battery so I don't think group loops are responsible.

I have tried adding various capacitors to the various lines and also looping ferite beads into the signal and power cables, no improvement. Capacitors to block DC component don't work.

I have tried an opto-isolator, but the output of the MP3 module isn't strong enough to drive it.

I am testing out using a small signal transformer from an old radio to isolate the MP3 module and the transmitter which is looking promising, but I worry about the ammount of power being taken from the Line out of the MP3 unit.

I can cure the problem immediately by pressing my thumb onto the back of the pre-amp OP-AMP. god knows if i's inductance, capacitance, or my bad sense of rhythm that cures it as I can't seem to re-create it with components.

The R feed originally went straight into the onboard DV camera which seemed to record it fine.

The pre-amp also picks up interference from my TV when too close. I suspect the cable to the camcorder is again acting as an antenna.

Anyone have any ideas?


05-13-2009, 11:10 AM
Try shielding your components that you don't want to receive the signal with aluminum or copper foil attached to ground.

05-14-2009, 06:18 AM
Hi, thanks, tried shielding all the different modules last night, no change to the interference.

I think that the signal is being picked up from the end of the audio cable at the transmitter. I believe that the transmitter is causing spikes on its power supply which is passing back down the audio cable and affecting the preamp.

I think i'll draw up a picture of the circuit as its getting quite difficult to describe.


05-14-2009, 07:18 AM

Here is a link to a picture of the units involved.

i just seem to be chasing this signal round and round in circles. I think I need a way to isolate the MP3 player from the transmitter. I think its really odd that the interference doesn't change volume when the master volume control is used, but does change volume when the pre-amp pre-set gain is adjusted. Probably very significant !!


05-20-2009, 07:18 AM
All sorted, I had added the preamp because the signal from the MP3 player to the main amp was too low. Between the MP3 and the pre-amp was a volume control that seems to have been wired up incorrectly. I connected the mp3 directly to the main amp and nearly blew out my ear-drums !!

One the volume control was wired ok and the pre-amp removed, no more interference. The AM reciever now picks up the signal ok from the main rover to blink the Manoi's mouth in time with the speech.

Just need to make up a small PCB now, get everything off the breadboards and build up some courage to start drilling holes in the Manoi's head for the LED mouth!!

Anyone if this would qualify for entry to the tutorial competition? and how to enter it?


05-27-2009, 06:09 AM
I have been working on a system to flash some LEDs on the humanoids mouth in time with speech from the chair unit.

Not as easy as it sounds. Lots of interference, ground loops, etc making life difficult. I seem to have made some progress last night in using a cheap AM 433mhz tx - rx module set. It worked well, but I got the occasional glitch which would flash the LEDs every now and again which was a bit too anoying to ignore.

I aded a couple of remote encoder chips to the circuit to encode and decode the signals to eliminate spurious signals which worked very well. I still had some odd interference which I traced back to a bit of hiss on the line from the MP3 player. An RC filter got rid fo enough of that so that I could use a transistor to set a threshold for enabling the transmitter.

Just need to move it all off the breadboards now and onto some veroboard tonight. Fingers crossed, I won't get any extra audio anomalies to deal with.

05-27-2009, 05:39 PM
Please excuse the mess, lots of late nights rummaging through trays of components.

Below is a quick and dirty video of the sound to light circuit in operation.

The sound you hear is coming from the robot chair on the left and being modified and transmitted to be shown on the LEDs on the breadboard. Tomorrow, all of that side of the circuit will be squeezed into Sir Kitt's head !!!

YouTube - sound to light test

Grand Robot Master
05-29-2009, 08:15 PM
WTF?!? the tux design is cool though;)

09-29-2009, 04:02 PM
I couldn't resist uploading this video from a recent show I attended with Sir Kitt.

Far too much Doctor Who on TV !!


09-29-2009, 04:07 PM
That is just too cool!

09-29-2009, 05:44 PM

09-29-2009, 05:57 PM

That is way too funny!

Good job on the robot thought. It looks really nice how it gets up from the chair. It really looks realistic.


10-02-2009, 02:13 AM

Tried to piece together some video from my robot (that wasn't subject to the non-disclosure agreement) to try to show some of its capabilities.

Got a bit carried away with Sony Vegas, feel like Spielberg now !!

YouTube - Sir kitt show