Challenge 2 - Can-grabber

"Bring those ten drink cans over here, and put them down the right way up."

This time it was our turn to set the challenge, and we went for a test of engineering and ingenuity:

The objective is to get ten empty drink cans back to the starting area. Two sizes of cans would be used (300ml and 500ml), and the cans must be stood upright in the scoring area.

The challenge ends when one robot has placed all ten cans in the scoring zone. Points will be awarded for each can standing the correct way up at this time.

Rules

We have chosen to allow rubber bands and other non-lego elements to build soft grippers.

The cans chosen are aluminium, so regrettably magnets (even lego magnets) will not be useful. No concessions will be made for cans unable to stand due to being crushed by robot grabbers.

Our solution

This challenge requires speed and accuracy. A conventional grabber robot will likely be doing the following:

• Drive to can
• Pick up can
• Turn around
• Drive to scoring zone
• Release can
• Turn around
• Repeat

If we can optimise this process, there's a good chance we will perform faster. Our solution was to remove the need for turning around.

I'm spinning around

A conventional robot must perform two 180 degree turns during each can-grab. Our robot's grabber will instead be designed to work on both sides of the robot. Our pickup process should then be:

• Drive to can
• Pick up can
• Reverse to scoring zone
• Deposit can
• Repeat

There are several ways to move the can from the front of our robot to the back. The easiest would be to place our grabber arm on a turrent ring and turn horizontally. This requires a grabber arm long enough to move around our tracks/wheels, which would be needlessly long and probably wobbly.

Another way is to swing the can over the top of the robot, but this results in the can being turned upside down. Adding a rotating wrist to the grabber arm would solve this problem, but introduces problems of its own: getting the hand lined up straight at each side, and dedicating one of our four control channels to it.

The solution was to connect the wrist and the shoulder action directly using gears. Thus, as the arm swings 180 degrees over the robot, the wrist turns 180 degrees and places the can down the same way up. As this was the biggest mechanical challenge, this is the part we made first.

It works a treat, and we're dead chuffed. Ice cream and chocolate sprinkles to celebrate.

Robot challenge 2 - results!

We met up at Blue Team's base, they have a nice long room, free of clutter and with a short carpet. Ideal battling territory.

The cans were laid out - we were one coke can short and had to sub in an extra relentless can. Blue team opted for us to deal with this, and we didn't mind - our grabber was happier with the taller cans.

A line was drawn, a short pause while Blue charged up their pneumatics, Blue's tracked robot was much faster, but their narrow jaws and jerky movement made it difficult to grab a can. Our machine crept up, made a grab and hit reverse while they were still lining up. First can down was ours, but Blue came zooming back, tracks clattering with ninja-like speed. Their pneumatic jaws opened and closed with a single switch-flick, but our reversible grip gave us a slight advantage.

Disaster struck when the Blue's robot blundered forward and knocked down several cans. Great care had to be taken at this stage as a nudge could send a can rolling and knock down others. Confused by which controller to release, I ket going forward when I only planned to move our grabber arm, which resulted in some of our own cans toppling. With no real steering, it was extremely unlikely that we could recover any more cans.

At this stage, we had 3 cans scored, to Blueteam's 2. Blue managed to pick up one of the two remaining upright cans, while we snuck around the back and grabbed the other. The race was on to place the last cans, resulting in a dead heat. 4 cans to 3, but technically, since we had picked up one of Blue Team's cans, it was disallowed, and a decider was called for.

One can, sudden death. It looked like Blue's speed would win this, but once again they found it difficult to make the pickup. Our wide jaw and slow speed made it easy, and we were once again reversing back to the line before the blues had made the grab. We were halfway back when they turned, ready to race back, it was going to be close, but some kind of controller issue sent the Blue's robot in an unsteady path, and we made the drop before they could cross the line. It was close, less than a second in it, but we had won.

CONCLUSIONS

Both machines seemed equally up to the task, but ultimately Blue's robot required skill to operate, while ours didn't even need to turn around. A perfect combination of the two designs would be: