Development of the reservoir refill system

 

      
There were a lot
      of challenges to getting this part of the system to work right.
      Circuit boards had to be designed and built, water pumps integrated
      and an IR modulation scheme to allow the robot to remotely turn
      on the receiver pump when it needed to. The ability for the robot
      to autonomously (all by itself) fill its own portable water tank
      when it gets low before heading out to water the plants without
      spilling a drop on the floor or making a mess is essential!

      
To accomplish this,
      the robots charging station was modified so that it not only
      has the spout at the top connected to the reservoir pump, but
      a IR receiver to generate the on/off signals to the water pump
      itself. The reservoir is a large square plastic bin, with an
      integrated lid which contains the tiny fountain pump at the bottom.
      Holes at the top allow the power cable to the pump and water
      outlet to pass through.

      
The system works
      quite well, right now the pump tends to fill the funnel on the
      robot a bit too fast and it can overflow if I dont back off on
      the flow with a valve or reduce the pump voltage a bit. For now
      to prove the concept I used a pinch valve on the rubber tubing
      from the pump to regulate the flow. It takes about 15 seconds
      to fill the tank from the half full position. More detail below!



      
Left: Front
      view, the current configuration. The filling funnel is seen on
      the right side, which has a yellow rubber hose leading into the
      top of the on board water tank, here which is at the full level.
      On the left is the pump outlet, which now is stiffened by a aluminum
      tube and puts out a 1/4" water stream pointing straight
      down, as to go into the plants filling funnels. The robot will
      not actually pour water directly on the plants - the robot will
      hit a funnel in the corner of each pot as to not risk damaging
      the delicate plants. The funnels on the pots will route the water
      right to the plants roots using small tubing. The dome protects
      the sensitive circuitry from dust, water and falling objects.
      (plant leaves can short the board out).
 

      
Left: Top
      view, showing the white fill funnel on the right and watering
      tube on the left.
 

      
Left: Here
      is what the water tank looks like when at the FULL level. its
      about half an inch from the top. There is an internal sensor
      to mark this location, as there also is one for overfilling it.
 

      
Left: The
      secret to making the robot have the ability to turn on the water
      pump in the big reservoir is here, the lower white LED is a 914nm
      infra red device of high power that transmits invisible digital
      data. The method I used modulates the 39khz carrier used in the
      IR remote sensor on the docking station with 1 khz. In other
      words - a unique signal that never occurs in nature or nearly
      any man made appliance. This and ONLY this signal can turn on
      the water pump on the reservoir to refill. Obviously with a 4
      gallon capacity, we must be careful to not dump the water out
      unless the robots funnel is under the spout!
 

      
Left: The
      start of the demo sequence. First I hook some jumpers up to the
      robots own water pump to drain the tank to empty. Normally, the
      robot would do this when watering the plants when on its "mission".
 

      
Left: To
      see the water level, the 4 red lamps light in sequence. Left
      is empty, next is FILL, followed by FULL and OVERFULL. Here,
      the FILL and OVERFILL is out since the water has not reached
      them with the tank half full.The robot uses this digital data
      to know when to fill its tank and when it is full and to turn
      off the fill pump on the reservoir.

 

      
Left: Robot
      is turned on simulating the point it will be approaching the
      dock when it has finished watering the plants.
 

      
Left: Now
      within 2 feet, it slowly drives forward - correcting its path
      so it will exactly connect with the charging plates on the dock.
      The red lamp in between the two brass charging plates is its
      aim point. It is actually homing in on the IR lamps inside the
      small clear bubble on the dock at the top.
 

      
Left: Dock
      achieved. The funnel is exactly below the fill faucet. Numerous
      tests show that after docking many times, the faucet is always
      within half an inch of the center of the funnel. This means the
      funnel is large enough to capture the stream.
 

      
Left: The
      robot next turns on the modulated beam and the water pump starts
      filling the funnel. It is set to never turn the pump on over
      30 seconds. Just in case the funnel is knocked off or something
      bad happens to its position.


      
Left: After
      the tank reports to the micro controller that it is full, the
      robot shuts off the tank pump.
 

      
Left: Full
      tank. The third LED is lit. The over fill LED is not...

 

      
Left: The
      display reads the tank is FULL and how many seconds it took to
      fill it. To do this, I simply kept track of how many cycles of
      modulated 39kc I sent. 14,000 cycles were 14 seconds. It takes
      1 mS (one thousandth of a second) to send one cycle. This is
      all generated in the master micro controller. (robots brain).
 

      
Left: Full
      tank with tiny 12v fountain pump inside. Ready to water the plants
      again.
 

      
Left: The
      front of the robot has a circuit board with a green LED at the
      top to indicate contact with the charging plates, and under it
      is a clear LED which transmits infra red data. Note the top of
      the IR LED is flattened with a file to spread the beam out to
      about 45 degrees width. The lens on a standard LED is much narrower,
      only about 15 degrees for this model. Those copper colored "wires"
      that contact the charge plates? Those are Guitar strings....
 

      

      
Left: Robot on
      charger/fill station parked waiting for night time. In the final
      version, the robot will look for daylight then wake up and go
      water the plants. When done, it spends the rest of the day back
      on the charger, now looking for night. Then it looks for daylight
      again...

      
This process is
      a bit complex, since it must take in account what happens if
      someone turns on the lights inside the house at night.



      
Left: The
      top of the reservoir tank. Holes on the upper right are for pump
      wires and tubing. A make shift pinch valve slows the flow of
      water for now.


      
Left: Inside
      the reservoir is quite spartan - a tiny fountain pump, same one
      as in the robot, and about 2 gallons of water. Cat litter bins
      are great!

 

      
Left: Electrical
      connections to pump and 18v power from the docking station. By
      making it easy to disconnect, we can fill up the reservoir fast
      with a minimal of effort.
 

      
Left: This
      is the micro controller board inside the docking station for
      receiving the modulated 39kc and converting it to pump on/off
      signals. The micro has its own 5v regulator, and the pump its
      own 12v regulator from the 18 v supply inside the dock. Everything
      is programmed in C. :)
 

      
Left: Schematic
      for pump receiver. Click to enlarge.
 

      
Movie 1
Left:
      a small movie of the robot docking and filling up. Yeah its tiny
      160 wide, but hey, you can hear the pumps turning on and off...

 

      
Interested in learning
      more on the development of this robot? Here are previous uploads
      - LATEST
      AT THE BOTTOM:

      
Previous Articles
1. Plant watering sensors
2. Cbot basic concepts and project definition
3.  Docking sensor, battery charging and FSM explained
4.  Development of the water tank, pump and feed mechanism

BACK TO ROBOT PAGES

HOME