We were invited to visit the LEGO Group at Denmark for 3 days at the end of September.  See our video summarizing the Moonbots Challenge, reaction of winning Moonbots, photos from Denmark, and the live Moonbots mission demonstration to the LEGO Group CEO.

In September 2010, the Moonbots award was put to good use for the FIRST Tech Challenge registration, thus began our rookie journey in the FTC division as team #4220, working with the unfamiliar metallic robot.

Landroids has conclude 3 successful seasons (counting 3 years and one month to be exact) as a FIRST LEGO League team, wrapping up our FLL career with the Moonbots grand price.  Our success has inspired closed to 100 students in our community and schools to join FIRST in 2010.  We have founded and expanded the Livingston Robotics Club in the last 3 years, which now has 12 teams in Jr. FLL/FLL/FTC this year.

On October 22, 2010, our team hosted a Moonbots Exhibition to 5,000 visitors as a part of the NJ Science and Engineering Festival (see photos).  At the same time, the first version of our FTC robot was participating in a RoboFest FTC scrimmage (see photos).  Talk about multi-tasking.

This Lunar Landroids website was created to document our Moonbots journey…which was a wild one, full of surprises!  It had completely took over our entire summer of 2010, but it was worthwhile with such chance of a life time.

Going forward, we will come back down to our official earth pad. Landroids’ progress will be posted on our  team website, with photos and videos posted on our Facebook.  If you happen to visit this Lunar Landroids website, please also follow our journey on traversing the obstacles in the “Get Over It” FTC season.  For now, signing off, Houston!

As a veteran FLL team, we thought we were very, even overly prepared for the live webcast with 6 cameras trained on our tiny robot, and upgraded our high speed internet connection. We had one camera and PC hard-wired for Skype, one camera to broadcast live to all of our friends and families, one fish-eye on the side, one top mounted for an overall view, one panning for the YouTube footage, and lastly, one Spy Cam on our robot.  This whole contraption took 2 days to setup and fine tuned.

At the exact time of our live broadcast on 8/24/10, everyone (past and new Landroids members, parents and grandparents alike) tuned in to watch, and everything went smoothly and quickly. We made one run, got 310 point, and happily signed off with Moonbots within less than 20 minutes.

However, our happy feeling did not last long.  When we checked the on-board video footage for submission, we found out that the on-board camera was out of memory, nothing was recorded! By trying to test and set up the time stamp properly in the morning, we used up the entire 1 GB memory in the Spy Cam.

By then, the team was out to lunch, literally.  With a couple panic emails, thankfully, Moonbots was kind enough to squeeze us in for another run.

Abandoning our sandwiches, we promptly reset the field.  The second run, the robot got 330 points, happy feeling was back! But the on-board video file had crashed, and we were happy no more, had to immediately call Moonbots again for another run.

After that, things deteriorated even faster!  Murphy’s Law struck us again and again!

By then, the robot battery level was low.  Once put in new batteries, the robot was too “hot”, we had to let it run wild for a bit to “burn off” some juice. While doing so, the robot flipped down the ramp a couple of times, parts came loose, and start to rub the chassis.

Finally, everything was readied, but the birds-eye camera was out of memory. More delay.

Then, the 1 GB on-board camera memory was full again, and files needed to be cleared out.  Time-out again while Moonbots patiently waited.

Just when we thought everything was ready, we started our robot run. Half way through, Moonbots lost Skype connection.  Sigh…..

Our team was ready to throw in the towel by then, but Moonbots reconnected the Skype, and did not give up on us. He was impressed with the first 2 runs, and wanted all the equipments to work well so that we could properly record the mission.

After combating Murphy’s army of gremlins, we finally got all of the equipments back on track, and went for a 3rd robot run, which got 330 points again.  Most importantly, the on-board video was recorded!  We now have a newfound respect for NASA trying to communicate with the rovers.

Watching our live webcast video, some people might ask, why did you leave one loop on the field? Well, we had a design that could potentially get 350 points, but it was not consistent enough, and there was not enough time to make it otherwise.  Beside, our team captain (aka my dad) promised me a bag of corn nuts for every 10 points we get over 300 points. So weighing the risks and probability, I decided to go for less points with a more solid design.

In the end, I didn’t get my revered 5 bags of corn nuts, but enough to last me till the end of the summer (probably destroying all my braces in the process).  The only problem was that when we went out afterwards to buy the corn nuts, there was no store that carried them. Murphy had the last laugh! So if anybody know where to get some corn nuts, please let me know.

With that, Landroids respectfully signed off.  Good luck to all Moonbots teams.  It has been a wild ride!

The final Phase 2 robot design followed our Phase 1 robot proposal and further develop the robot arm and refine the hopper. Initially, the design was to spear and flip loops into a top hopper on the robot. However, as we started to consider collecting loops placed in any orientation, the robot arm design soon evolved into a complicated claw assembly. Eventually, the claw robot has to be discarded in order to complete the missions efficiently.  We felt that we had a good claw design, but we also learned that sometimes, when KISS is necessary, we had to sacrafice our pride and sentiment in order to move forward.

Our final robot is a much simpler fork flipper robot for mechanical consistency. The Flipper Robot Design was divided into a couple major components:

End Effectors: A flipping fork and a top mounted expandable hopper
Motors:  3 NXT motors
Drivetrain: Rear wheel drive, 2 wide flat rubber tires, 2 tank treads, 2 front skis with rollers and wheels
Sensors: Rotation, 1 compass, 1 EOPD, and 2 ultrasonic sensors
Accessories: Key Fob Spycam with the built-in time stamp
Programming: LabView

End Effectors

Robot in Base

The robot flipper is a straight arm consisting of beams with a rotating fork at the end. The fork has a pair of prongs; its width is wide enough to grab both loops at once but narrow enough to fit within the A4 robot size limit.  The fork is parallel with the ground when the arm is down to spear the loops. When the arm is raised, the fork rotates back by gravity and drops the speared loops into the top hopper. When the fork is lowered again, there is a rubber band to pull the fork back to be parallel to the ground.

The expandable hopper sits on top of the arm motor to collect loops. It mainly consists of axles to reduce weight but maintain rigidity.  Both sides of the hopper can spring out like wings to a larger funnel when the arm is lowered after the robot deploys from base.

Spearing loops

Motors

Arm motor and sensors

Two NXT motors are used for the drivetrain, and the third NXT motor sits on top of the drive train to power the arm through a 2:1 gear ratio for more power and accuracy.  The gear train is sandwiched between beams so they do not move apart or spin freely from one another, which was a problem encountered when the arm is overused or the gears were under stress.

Drivetrain

Robot front skis assembly

The robot uses a rear-wheel-drive system and two tank treads on the underside. This unique design was developed in Phase 1 and employs the advantages of both drivetrain systems without a very complicated gear train or suspension.  The front is supported by two small (red) plastic wheels that have curved edges to make turning smoother.

There are two skis with embedded wheels for gliding over the ridge.  The skis are suspended off of the ground to avoid dragging over on the surface. The back wheels are a pair of medium size tires for maneuverability and traction; they are the main driving force for navigation (forwards, turn, backwards).

Tank treads over ridge

Tank treads on underside

In addition, two tank treads tucked between the wheels/tires/skis along the underside of the robot. Each tank tread is powered separately using a different NXT drive motor.  When the front wheels and the skis are over the ridge and the back tires are slightly suspended in the air, the tank treads can add power and pull the robot over the ridge.  The main structure of the robot is shaped like an upside-down U for ridge traversing.  The tank treads span the middle of this U.  For structural strength and compactness, the NXT sits in the middle of the drivetrain, therefore adding strength and a low center of gravity.

Sensors

The sensors used on this robot are:

1EOPD, 2 ultrasounds and 1 compass sensor + Key Fob Spycam

• The built-in NXT rotation sensors for basic movement (forwards and backwards);
• A compass sensor to tell the robots angle compared to the beginning at the lunar base (it takes a calibration);
• Two ultrasonic sensors for wall-following and distance measurement (one facing the left side and one facing the front);
• An EOPD sensor in the front mounted between the two skis for landmark detection (crater ridges, moon rocks, and peak of eternal light, etc).

Programming

The initial programming was done in NXT-G; however, the file got so large that the compiler crashed, and the program were accidentally deleted one week prior to the deadline. The new program was rewritten in LabView (first time ever using LabView for robot programming), which can handle a more complex program, faster and smaller compiled code.

The Claw Robot Design was divided into couple different components:

End Effectors: A double claw assembly, cable arm, fold-up basket hopper
Motors: 2 Power Function and 3 NXT motors
Drivetrain: Rear wheel drive, 2 wide flat rubber tires, 2 tank treads, 2 front skis with rollers
Sensors: EOPD, compass, ultrasonic sensor, IR Link
Accessories: Video Camera and counterweight

End Effectors

A four-way claw as envisioned in Moonbots Phase 1 was built to collect loops placed in any orientation. In order to lift the massive arm, a pulley system was used.  This made it easier to raise the arm but too slow to even think of getting more than 4 loops in the alloted 3 minutes.  To speed up the process, two claws were used to grab two loops at oncep. The claws were powered by a single motor driving both claws using chains, and the claw assembly can also rotate by another motor to grab loops when robot was coming in at an angle and to drop the loops into a foldable hopper below.

Double Claw assembly

With such a large robot, the main concern was accuracy. To avoid any instances where the robot could become disoriented, a long arm was used to suspend the claws over the crater, so that the robot wouldn’t need to enter any craters. The arm was attached to the base of a mast which served as the anchor points for the 6:1 pulley, limiter strings, load reduction rubber band, sensor mounting points, and a very convenient handle to carry the robot. To provide more lifting power, the NXT motor was reduced to a 2:1 ratio though a gear train which pulled the line running through the 6:1 pulley assembly. If this pulley design were to be finalized, we had a messy roll of LEGO lines at hand to swap out all temporary lines.

Pulley arm design

Also, if the double claws are placed wide enough to grab both loops, then the hopper to catch the loops will be wider than width of a piece of A4 paper.  Therefore, a foldable collection basket was used to accommodate the width of the double claws.  The basket would drop down as the robot arm unfolded after the robot left base. The basket was anchored to the front of the robot and clipped to the middle of the arm so the entire package when folded is less than the A4 size required.  The tip of the basket also provided a forward mounting location for the EOPD sensor.

Fold out loop basket

With so much weight in front of the robot, it was very front heavy and easily tipped when coming down the ramp or over the ridge, so a folding counter balancing tail was added. By mounting the Power Function battery pack on the tail and a stack of large LEGO wheels that flip out, we managed to achieve a 50/50 front/rear weight bias on the wheels of the robot.

Counter weight with Power Function battery

Motors

Two NXT motors were used for the drive train, one NXT motor was for lifting the arm.  Two Power Function Motors were used to control both the opening and closing of the claws and the rotation of the claws, while the collection basket and robot counter tail would unfold by gravity.

Drivetrain

Due to the weight of all the end effectors, the drive axles were bending under the weight of the robot.  To create additional support for the axel,  support beams were connected to the outside and inside of the drive wheels hence taking some load off of the axles. Double tank treads were used on the underside to increase traction and power when climbing over the ridge. To make it easier for the robot to turn, small plastic wheels were used for the main robot navigation while skis were mounted in the front to glide over the ridge.  This drive train design pretty much adhere to our Phase 1 concept.

Robot folds up in base

Navigation

Navigation was provided by a combination of Compass, Ultrasound, EOPD and NXT motor rotation sensors. Rotation sensors provided basic distance information, Compass sensors gave heading information, and Ultrasound and EOPD together were used in wall-following and landmark detection. The last sensors port was used up for the Power Function IR Link to control the Power Function motors.

Design Pros and Cons

Pro: Able to pick up loops at any directions
Con: Too slow and too massive, hard to balance and manuver