I am building a tracked robot which will need to be able to find its charging station. I have two ideas for how this might work.

1. On the charging station, an IR LED flashing at, perhaps, 1kHz. On the robot, an IR photo-transistor in a tube to make it more directional. Then move the robot to find maximum received 1kHz signal. The MCU (Pic) would need to remove possinle interference from fluorescent or LED lamps and sunshine. Perhaps just a high pass filter? How to implement this?

2. On the charging station, two LEDs flashing at different rates, arranged horizontally with a vertical screen between them. Now the photo-transistor will receive one frequency if to the left of the station and a different frequency if to the right. The MCU will need to be able to reject interference as above but now also needs to be able to distinguish the two frequencies. How best to implement this?

 

The IR approach is the one used by the Roomba and its copies. You might look for a dead or very old Roomba to canabalize.

As far as I understand it, it is a matter of successive approximation. More than one sensor is used so the robot can find the charger even if it is not directly in front. The Roomba keeps a map for navigation, so the sensors are restricted to the front part but you might need them all around.

 

I have an original Roomba, and to turns in a circle to find the station, so only front sensors. Newer ones may be different. Same with the map, I read about it back then and it does not use a map, they tried but it proved to be difficult and no better than random patterns, which they used instead, but that was many years ago.

 

As it is tracked it can spin on the spot so i think only one sensor will be needed.
I hope this thing will build a map so it should be able to get back roughly into position. That's the plan anyway. However it needs to get it just right so the charging connectors, err, connect.

 

As it is tracked it can spin on the spot so i think only one sensor will be needed.
I hope this thing will build a map so it should be able to get back roughly into position. That's the plan anyway. However it needs to get it just right so the charging connectors, err, connect.

Hi,

I did an IR project way back when and i got good results with a multistage RC high pass filter. I might have used 5 stages I don't remember that well now. That filters out ambient lighting like you were thinking.
One of the key points is the level of brightness vs the distance, and that is coupled with the direction and beam width of the transmitting IR signal. A focusing system helps here a lot using lenses.

As to your connectors, you can use rare earth magnetic connectors. They can almost find themselves when brought within about maybe 1/2 inch, so the positioning would not have to be super-duper accurate, just within reason. You might get away with just four rare earth magnets two for each pole, one on the station one on the moving unit. When they connect together, they can conduct current. They make magnetic connectors too, but you'd have to find some appropriate for your current and voltage and size might be an issue.

There might be other ideas as to the navigation too.
For example, instead of just a single or double frequency the station could send out digital codes.
Also, if you don't mind another unit or two, you could have other beacons around the room that help with the navigation. They could all send out different codes.
Duplex communications might help in some way.
Mirrors can help direct one or more signals to different places. For example, if you had four mirrors set up in each of the four ceiling corners the station could send signals to all four. Since each signal code would be different, the moving unit could detect which one it was closest to and make adjustments to the direction of travel accordingly. if you dont like mirrors perhaps tiny secondary transmitters linked to the station. They would have to be continuously powered though.

The only thing about IR is that if something gets in the way how does it know where to go. You might have to implement a "search for the signal" algorithm so the unit can move around until it finally gets a signal unless you use several beacons of some kind.

 

I have an original Roomba, and to turns in a circle to find the station, so only front sensors. Newer ones may be different. Same with the map, I read about it back then and it does not use a map, they tried but it proved to be difficult and no better than random patterns, which they used instead, but that was many years ago.

It definitely makes a map now. You can view it on their mobile app.

 

As to your connectors, you can use rare earth magnetic connectors. They can almost find themselves when brought within about maybe 1/2 inch, so the positioning would not have to be super-duper accurate, just within reason. You might get away with just four rare earth magnets two for each pole, one on the station one on the moving unit. When they connect together, they can conduct current. They make magnetic connectors too, but you'd have to find some appropriate for your current and voltage and size might be an issue.

I have some magneric connectors that would be fine electrically but the problem I forsee is that they are very strong and I doubt the tracks would have enough traction to be able to be able to pull them apart. I also got some inductive charger things, but they need to be pretty well aligned and have fairly limited current ability.

My plan at the moment is to have spring test contacts on the station and I have got the (gold plated) contacts off a 1/4" jack socket mounted to match the spring pins.

I did an IR project way back when and i got good results with a multistage RC high pass filter. I might have used 5 stages I don't remember that well now. That filters out ambient lighting like you were thinking.
One of the key points is the level of brightness vs the distance, and that is coupled with the direction and beam width of the transmitting IR signal. A focusing system helps here a lot using lenses.

OK, thanks for that information. It is supposed to be able to make a map and navigate from place to place with reasonable accuracy so it should be able to get fairly close to the station, just needing the IR for the fine positioning. I plan to cheat slightly and have a tapered entrance to the station to do the final alignment.

 

The robot vacuums use simple plated sliding contacts with the springy part on the robot. You could fabricate something similar using the nickel strips used for 18650 battery packs.

If the charger has a pair of channels to direct the robot in and align the contacts it should be reliable.