Considerations for self charging robot

[izua]

I'm planning to design a self-charging platform and electronics, but I want more advice.

The first issue is - how are the accus connected and how will they be charged?
If i'll put them in a battery holder with 4 slots, they'll be in series and I'll have to charge them in series. Bad idea for NiMH. If I'll put the in individual holders, I'll need some way of toggling between putting them all in series (for motor powering) and individual access (for charging). There's also the issue of powering (and charging!) the electronics that does the toggling.
Charging, I think that's another problem. It could be done with individual touching connectors for each battery, or it can be done simply with a power connector. In the first case, the robot will be lighter, while in the second case, although the higher versatility, I'll also have to build a charger.

Another issue is the connector. I was thinking that thick twisted wire as a whisker might do. I could leave it suspended above ground level, so the charging platform will have rails for guiding. I can't see any other option for this.

The last question is, naturally, about locating the charging station. There's really no use in having the robot self-charge, IMO, if it can't go to the charge station by itself. Using infrared beacons might do, but I'm thinking about better options. The problem of IR beacons is that it needs a well designed protocol if you want to have multiple beacons.

Any ideas appreciated.
thanks
-izua

[AndrewM]

If i'll put them in a battery holder with 4 slots, they'll be in series and I'll have to charge them in series. Bad idea for NiMH.

Why do you consider this a bad idea? 

[maverick monk]

there is nothing wrong with chargin nimh in serise, just make sure your charging at the same voltage as the total pack (4 cells being 5v.) charge time is battery mah/charge rate in mah. nimh battery packs are charged in serise. its perfectly fine

[cooldog]

The last question is, naturally, about locating the charging station. There's really no use in having the robot self-charge, IMO, if it can't go to the charge station by itself. Using infrared beacons might do, but I'm thinking about better options. The problem of IR beacons is that it needs a well designed protocol if you want to have multiple beacons.

you can just cover the station with ir leds and have the robot run a photovore algoratum using ir range finders instead of photocells

[Ro-Bot-X]

RC cars have NiMH battery packs that are a number of single cells connected in series. Same for the power tools except they are using NiCad cells. They have a special charger that will charge the whole pack like Maverick Monk has explained. To protect your electronics add a diode in series with the positive lead of the battery pack and connect the charger before the diode. It is also possibe to have the electronics powered while charging, but in this case the charger should supply the amperage needed for charging AND the amperage needed by the electronics, without motors of course. How to separate motors from the electronics? Well, you might use a relay to switch off the power rail for servos and drive motors while charging. The relay will be powered only when charging. To connect to the station use a ground plate located on the floor and a wall plate located on the face of the station (or create a ceiling like for the bumper cars) and a couple of springy antenas. You can even use the same relay to swing the antenas in charging position.

As for station localisation, you can use several methods. One is using IR beaming like you mentioned, another is map localisation then use the distance sensors to center on it, or use the method mentioned by Cooldog.

[izua]

I believe niMH charging in series is bad because i can't control the cells individually. This way, if during operation a cell drops below 0, i'm not sure charging it in series will do any good.

There's also the problem, if using 4 (or individual battery holders of how do I connect them to be measured individually?

Using an analog mux might give more errors, to switch the reference point for an ADC channel. But here another factor arises, as i sample the voltage of different batteries, while they are wired in series, the reference voltage will rise with respect to the common ground. at the fifth battery, i might already be at 1.2 * 5, 6 volts. So, a method for sampling accus individually might be needed, too.

covering the station with ir leds sounds fine, except that, that's the only beacon i can have. put two beacons like this in a room, and the robot can't find its way anymore to one of them.

[Admin]

I've seen a few online with documenation, just browse around:
http://www.google.com/search?hl=en&safe=off&rls=DGUS%2CDGUS%3A2006-11%2CDGUS%3Aen&q=self+recharging+robot+schematic

http://www.schursastrophotography.com/robotics/dockinglogicdocking.html

[Webbot]

Base station location. Another spin would be a line follower - but you may not want to paint a line on your carpet. So other alternative is that the station is connected to a wire laid under the carpet. Place some coils under the robot to detect the current in the wire. Then program as per the line follower. The same wire could go via multiple stations or each station could cover its own private area. Just be careful that the wire doesn't get hot !!