Roboscout Revival - by Paul Backhouse
There should have been three.
But tragedy struck when a near mint, boxed, example being shipped over from the US was mercilessly intercepted and crushed by US Customs for being in possession of a battery. Sadly one less Roboscout on the planet. I had a deal of sympathy for the seller. Roboscout’s large SLA is difficult to remove, even after detachment of 15 miniature Molex PCB plugs and extraction of the main board.
My Roboscouts were made in 2000 by Sharper Image Design. I’m sure I’m not the only one to have spent many a happy head-scratching hour trying to reverse engineer Roboscout’s convoluted PIC-based mainboard - without success. One common fail seems to be a chip which switches between charging and onboard power.
Roger needs a new semi-autonomous brain.
This post is about onboard power, speech and movement. I have more work to do on Roger’s voice recognition and sensory function including Roboscout’s hidden surveillance camera and infra-red sensors.
And a final cautionary note - this post contains my own “venerable” solutions, which some start-of-art guys may find upsetting….
The first step was to scrap Roger’s SLA (lead acid) lump, and replace it with 2 x 10 x 2400mA rechargeable NiMH cells and an LED battery monitor screen.
The space vacated by the SLA is now occupied by two Arduino Megas. A rectangular cut-out in Roger’s base gives access to their USB programming ports.
For Roger, I decided not to go down my usual “SP0256 Narrator” route. I opted instead for a probably equally obsolete WTV020SD16P card, which, rather cleverly, can randomly access Octally indexed .AD4 audio files (speech phrases) stored on an ordinary SD card. Roger currently has a vocabulary of 178 words and phrases on one card. Here, a WTV020SD16P card undergoes a quick test from a Uno, without an amplifier. So sound right up🔊…
Speech generation from .AD4 files had 4 stages. This seems daunting, but was fairly straightforward. Here’s how I created Roger’s natural speech vocabulary:-
Format (overwrite) the SD cards using dedicated SDcardformatter from sdcard.org. Note, not Windows format.
On the FreeTTS.com website, select the language and voice, and input the required text to create .MP3 files which can then be downloaded.
Use onlineconverter.com/mp3-to-wav site to convert each MP3 file to a WAV file.
Use AD4convertor to convert .WAV to .AD4. I installed “AD4convertor” a couple of years ago as a zipped app. It still seems to be available on several sites. You execute it at the command prompt, and it looks like this:- AD4convertor -E4 0000.WAV where “0000” will be assigned as the index number for this word or phrase on the SD card.
I can hear the screams from here. There are *much* easier ways to create a natural robot voice…
Probably… What I like about this method is the facility to have large randomly accessible vocabularies on removable, swappable SD cards.
One final pitfall for the unwary. WTV020SD16P expects 3.5 volt data, not 5 volt data. This is easily accommodated by including a 470ohm resistor in each data line.
Here was lying in wait a further conundrum. Roboscout’s motor function is driven by brushed electric motors, with simple photointerruptor feedback. Body and limb construction don’t easily lend themselves to servo conversion, which would have been the ideal way to convert Roger to Arduino control.
The best option was to interface digital outputs from two Arduino Megas to12 volt relays via 2N3904 transistors, and to introduce potentiometers for analogue feedback where this could improve on the photointerruptors. I was able to reuse ten of the existing Molex-type motor-feedback connections to new PCB plugs on two relay boards. Each Mega controlling 10 relays on its board.
I decided to use separate relays for left/right, up/down, clockwise/anticlockwise functions. I could have used DPDT relays, but these would have needed separate provision for the “off” state. These separate relays also simplify interfacing the Megas’ digital channels (eg: one digital output for up, another for down.)
Roger’s first brain motor function test, sound right up…🔊
The row of 36 1.8mm LEDs grouped in fours has become my signature interface!
After a little cosmetic work on his eye mood colours and mouth LEDs….
Sound right up 🔊…
You might like some of my other articles:-
ncG1vNJzZmiokaq5o63CpJ%2BoraOae7TBwayrmpubY7CwuY6pZqunkqTApLvUrWSrnaaew6K4