2012 Lunabotics | Page 1
You may like to read 2011 Lunabotics first to explain more about the competition and to see how we preformed in the 2010-2011 NASA Lunabotics Mining Challenge. Our 2012 rules from NASA, Systems Engineering Paper, and Slide Presentation can be downloaded here:
- 2012 NASA Lunabotics Rules (850KB PDF)
- 2012 Systems Engineering Paper (8.4MB PDF)
- 2012 Lunabotics Slide Presentation (5.5MB PDF)
My experience in the 2011 lunabotics competition really helped me prepare for the 2012 cycle. Some of our team members left the team due to graduation or moving. The success of the 2011 Alabama Lunabotics team made it easy to recruit new members. We also partnered with the local community college, Shelton State.
I was elected team lead for the 2011-2012 season.
Once again, SolidWorks was chosen to 3D model the robot during the design phase. As a team, we decided to re-design the robot from scratch. This was partly because of rule changes and partly due to design flaws that could hinder us. We had learned a tremendous amount from the previous year's modular design which we built on top of.
Some of the issues with the 2011 base were:
- The drive motors did not provide enough power to the wheels.
- The new rule changes made the weight of the robot very critical. The use of 80/20 was not an optimized system and could be improved upon.
- The electronic components for the modules were mounted to the base. This limits the modules built on top of it greatly.
- The usb webcams through the Phidgets SBC1 board hurt us last competition. It took ~10 seconds to switch between the cameras which was wasted time.
Taking these things into account, we were able to design a new base.
The new base:
- Much larger weight capacity
- Smaller footprint to better cope with the new dimension requirements
- Lighter overall weight
- Larger drive motors for more torque at the wheel (Several hundred ft/lbs)
- Only contains electronics for the base functionality. A single connector that provides power and communication to the module allows each module to contain the needed electronic components. This also reduces weight compared to the 2011 Lunabot.
- Much larger battery current and capacity ratings
- Still able to interface with modules from the previous year
Redesigning the robot allowed us to use the 2011 lunabot in our outreaches and other demonstrations. This was great since not only could we start outreaches immediately, but we also did not have to worry about transporting the new robot while it was still under construction. Here is an image of the 2012 robot base module with the wheels from the previous year and no electronics box.
We did carry over the sweeping wheel design while improving on the idea. Instead of two linear actuators and linkages controlling the wheel positions, we designed our own worm/worm gear rotary actuators -- one for each wheel, giving us the ability to place each wheel in any position independent from the other wheels. From a controls side this ability gave us tremendous options. For instance, if we lost power to one wheel or it locked up, we could actuate that wheel and walk the robot to safety. The below image shows our robot in the home position, driving sideways, zero-point turning, and random positions of each wheel.
Once again, 80/20 was chosen for the upper portion of the base module. We could have used other material, but the slots in the 80/20 allows non-destructive changes to be made if we needed to adjust our interface to future modules. A 1/4" 6061 aluminum plate was used for the base of the robot. The plate was machined to be lighter without losing the necessary strength.
Not only did we upgrade the motors from the previous year, but we also improved the gearboxes. A move from aluminum ring gears to all hardened-steel gear boxes increased our power capability. The gear reduction we needed (256:1) was not available in this gear box, so we had to modify and machine two 64:1 gearboxes together. Here is a mock up of one sweeping wheel corner and a 64:1 gearbox. The worm/gear on top controls the position of the wheel relative to the robot base.
This image shows that with the right type of wheel, positions beyond 0 and 90 degrees are capable. This could be used to drive the robot at angles accounting for wheel slippage or drag.
Being engineers, we did believe the weight capacity numbers from our designs, but testing is always fun. Here is a simple static load test.
Solid as a rock with two people standing on it.
The improved battery current and capacity abilities came in the form of 5Ah 5S Li-Po batteries. The included connectors were not up for what we needed, so we had to upgrade them.
The base could hold 4 of these at a time, giving us:
- 18.5v main rail voltage
- 20Ah battery capacity
- 1,200 amps peak current
- 600 amps constant current
The electronics box for this base was a tremendous challenge. A large number of electronic components, many high current ones, had to fit inside of a small 6"x5"x5.5" enclosure. We mounted our circuit breakers to the top of the box, giving us easy access in the event of a trip. The 4 150amp circuit breakers cut our 1,200 amp peak current draw down to 600 amps peak. This should still be more than enough for the current robot needs.
This enclosure contained:
- A 500 amp shunt resistor
- 1,200 amp relay
- 2X 2 channel 120 amp peak motor controllers (Drive motors)
- 2x 2 channel 50 amp peak motor controllers (Sweeping motors)
- Wireless router running DD-WRT
- Phidgets SBC 2 (Onboard computer)
- 5v and 12v voltage regulators
- All of the necessary wiring for the above systems (Including the main 6ga and 8ga wire)
It was not pretty when completed, nor was it the most internally RF friendly device, but it was more than enough to meet our needs.
With the electronics box complete, we started testing the base to probe for any weaknesses that needed improving. We visited the local volley ball pit, again, to give us some soft, uneven terrain to drive on. As mentioned in 2011 Lunabotics, the competition uses a lunar regolith (soil) simulant called BP-1, which consists of the finest powder that you can imagine mixed with small penny-sized gravel. Once compacted, the top layer is very fluffy and the subsurface material is extremely hard. Take note in the video below, around the 3:00 mark you can see the obvious difference between skid-steering or our zero-point steering. Sand is more supportive compared to the BP-1 at the competition, and you can still see how we tear it up and dig in when skid steering.
Due to our awesome team member Jason and his connections, we acquired some BP-1 of our own. Over a ton of it actually. Unfortunately, it was not quite enough to drive the full robot around in, but we could use it to test digging implements. I designed a wooden enclosure to build around the box that would allow us to mount digging heads and test various styles of excavation in actual BP-1 simulant. Unfortunately, we did not have enough time left in the year to be able to use this effectively, so we saved the money.
With the base working well, the focus moved to finishing the module design. The team proposed many module ideas, each of which we drew up in SolidWorks to compare their true benefits and pitfalls.
The chosen module consisted of:
- Dual bucket wheel excavator (BWE), industry inspired
- Cleated conveyor belt delivery from collection BWE to onboard storage hopper
- Screw conveyor mounted to the bottom of the hopper to move the regolith to the back of the hopper
- Cleated conveyor belt delivery from the hopper to the competition collection bin.
In order to meet our starting dimension requirements, we had to make the loading conveyor belt fit inside of the hopper. Below, you will see the overall design, as well as the stored, partially out, and digging positions of the BWE arm. This arm also allows us to control the digging depth of our robot.
This design, when partnered with our sweeping wheels, allows for massive regolith collection. We can essentially drive sideways, digging a trough until our hopper is full.
We can collect regolith in any of our driving directions:
The hopper can store over ~125kg (275lbs) of regolith.
Concurrent work was happening on the software while we were building the base. We had one team member dedicated to coding the robot backend while another coded the GUI front end. They decided on a protocol together and it really turned out great. The software was quite difficult due to the various modes of our robot and the large number of possibilities that could be done with it. We decided as a team to use two robot operators instead of one. One operator for the base and another for the module. Each operator used an XBOX 360 controller to interface with the software. Here is an animation of the development of our software repository.
As the BWE design finished up, we started ordering parts for assembly. The assembly of this module was more labor intensive than the previous year. We had several aluminum plates that had to be machined and other parts that needed cutting and welding.
One of the few things that we purchased, instead of making ourselves, was the screw conveyor. It shipped with a 1/4" still square tubing center (pictured left) which we replaced with aluminum to save weight (about 15lbs lighter)
To really show you the difference between the gearboxes between the 2011 and 2012 robot, take a look below. The gearbox on top drives the 2012 robot (and the Bucket wheels) and the bottom motor drove the 2011 robot (and conveyor belts for the 2012 robot). The motors pictured provide ~%300 more torque than the ones used in 2011.
Getting a little dirty with some early testing. The motors in the picture above can be seen near the end of the bucket wheel arm.
The bucket wheels were constructed by hand. They are some of the few non-aluminum components on the robot. We welded cut 3" steel pipe pieces between sheet steel to make them.
Once completed, the wheels were mounted to the BWE arm. With the robot starting to take shape, we added the light cloth walls of the hopper. The cloth is very tough, but lighter than if we had used fiberglass or other available options.
An early attempt to seal the drive motors used custom cuben fiber boots to keep dust and dirt out. Unfortunately, they had to be replaced due to a need for more airflow.
We instead chose to install filters onto the motors to keep large particles out of the motors. Next, we sealed the entire base to keep the internal area as clean as possible. This was rather difficult with our sweeping wheels, but some leftover hopper cloth worked great. Not pictured are the top velcro flaps that were sealed with tape during competition runs. You can also see the module power/communications connector and physical connector rails. The rails allowed us to attach the modules with just 4 pins, while the electrical connector provided power, USB, ethernet, power switch, and status light interfaces to the base.
Staying weight conscious, we decided that we needed new wheels. The old wheels consisted of 10" pvc pipe, 1/4" aluminum plates, 3/4" angle aluminum cleates, and 1/4" hubs. They were effective, but heavy. Several teammembers submitted wheel designs, but we settled on the design featured below. Here, you can see one of the stress tests completed to ensure that the wheels would hold-up under various conditions.
More cloth was added to mitigate dust. At the competition we are judged on how well we are able to handle the dust created by our lunabots. The completed system would encase the conveyor belts and hopper in this cloth.
Here, you can see two of the three PTZ IP cameras on the robot. We also mounted a static usb webcam under the bucket wheel to watch the digging process more closely.
Another video showing more of the building process and testing.
The competition week grew closer and we still only had one tested module. We designed our base to accept the modules from the previous year, but we had not been able to test them mated with the new base yet. Taking into account the new rules which deduct points for every kilogram that we weigh, we took the 2011 front-end loader and made it as light as possible without losing functionality. Here is what we ended up with.
The bucket wheel excavator mounted to the base weighs right at our maximum allowed weight of 80kg. The front-end loader mounted to the base only weighs about 45kg. With these two very different modules, we could show up at the competition and essentially have two different robots. Here is the front-end loader (FEL) module mounted to the base, holding the wheels from the previous year.
Our maximum payload in the FEL was about ~10kg -- not because of lifting power, but because the center of gravity would shift far enough to tip the robot. Our sweeping wheel design allowed us to overcome this by sweeping our wheels into 90 degree mode (shown above). In this mode, we could easily carry over ~10kg, which meant that in the competition we could qualify to win with only one scoop.
- DIY $25 Steadicam
- Stereo Cooler
- Portable Potato Cannon
- Microcontroller Launcher
- High Altitude Glider
- Hard Drive Speakers
- Portable PC
- Java Game
- La Fonera Wifi Routers
- NES Optical Mouse
- Floppy Drive Camera
- Potato Cannon
Check back for more projects and updates!
All of these projects are self funded. Donations are appreciated. Thank you.