more challenges.. more money..

July 16, 2010 - 2:39 pm 1 Comment

NASA announced 3 new Centennial Challenges…

The Nano-Satellite Launch Challenge is to place a small satellite into Earth orbit, twice in one week, with a prize of $2 million. The goals of this challenge are to stimulate innovations in low-cost launch technology and encourage creation of commercial nano-satellite delivery services.

The Night Rover Challenge is to demonstrate a solar-powered exploration vehicle that can operate in darkness using its own stored energy. The prize purse is $1.5 million. The objective is to stimulate innovations in energy storage technologies of value in extreme space environments, such as the surface of the moon, or for electric vehicles and renewable energy systems on Earth.

The Sample Return Robot Challenge is to demonstrate a robot that can locate and retrieve geologic samples from wide and varied terrain without human control. This challenge has a prize purse of $1.5 million. The objectives are to encourage innovations in automatic navigation and robotic manipulator technologies.

More information can be found here. There is a very large possibility that we will compete in both the Night Rover and Sample Return challenges. First dibs on calling our robot “Lady of the Night”.

2009 Event post summary

December 2, 2009 - 8:26 am 1 Comment

2009 Regolith Excavation (and Networking) Challenge

Much More Than We Anticipated, Less Than We Deserved


Sorry for this being published late, I have been catching up on all the things I have let slide for the past two years.

First and foremost, I need to thank the contributors to the project. As always, the emotional and financial support of my wife Valerie enables me to do these crazy things. The software crew (Denny Gudea and Aaron Shumate) created a NASA worthy control interface and solved the “networking with delay” problem that prevented the vast majority of excavators from working. As always, Jim Hart provided aerospace quality work essential to make a system like this reliable. Greg Shreve, along with Jim, creatively solved a number of design issues critical to the function of the excavator. On the mechanical side, Ken Sterk’s help in manufacturing and assembly enabled the completion of the Brute Force Digger. Rob Sweeney’s algorithms for the IR sensors on BFD 3 enabled a method to measure the accumulated Regolith where before, the noise made the sensors unusable. Every one of the team members made significant contributions that enabled us to field two highly competitive excavators.


Before leaving for the 2009 Regolith Excavation Challenge, our team had done considerable research on the capabilities of other teams. The competing teams broke into three categories: teams with mining capabilities much greater than ours, teams with mining capabilities around the same magnitude as ours (>200 kg) and dark horses (no information). With apologies to the dark horses, the list broke out as:

a)     Greater

Paul’s Robotics  from Worcester, MA
Team Waldbaum from Sunnyvale, CA

b)    About the same

C2 Robotics from San Luis Obispo, CA,
Invading Huns from Palm Bay, FL,
Moon Diggers A from San Francisco, CA
Moon Diggers B from San Francisco, CA
Technology Ranch from Arroyo Grande, CA

c)     Dark horses -

Auric Designs Orlando, FL
Colorado School of Mines Golden, CO
E-REX Little Rock, AR
Green Cheese Solutions Hudson, WI
Innovation Island Southwest Harbor, ME
Laurentian Rock Glendale, CA
Lunar Arc Irvine, CA
McGill Lunarex Knoxville, TN
Next Step Robotics Houston, TX
Ockham’s Razor Arroyo Grande, CA
Sandstorm + North Royalton, OH
Seraphim Aerospace Systems Moshannon, PA
Top Hat Robotics West Lafayette, IN
UBC TREAD Robotics Vancouver

Looking at this list and making simple math assumptions based on previous contests and observed maturity, it is easy to calculate where we should expect to end up. On average, about 1.5 teams out of category A would work and surpass us. In category B, About 2.5 teams would work and 1.25 surpass us. In category C, about 5 teams were expected to work and 1.5 teams surpass us.

When you add all of the contributions, we expected to average about fifth place. Our experience and extra practices were probably good for one more place. That put us in forth and fifth. Without too much of a statistical anomaly, third was in reach. We would need a major string of failures to get second and an outbreak of H1N1 to get first. That was our expectation going into the event.

Arrival and Robot Checkout

We arrived in San Jose Thursday, October 15th. Like most teams, we had been pushing hard for the past couple of months and everyone arrived needing a good night’s sleep. I fell into bed at 8:30 pm and stayed there until 7:30 am the next day. We met the rest of the team at 9:30 am at the gate in front of Moffett field. Of course, nobody seemed to know who we were, where we were to go and what we were to do. Eventually, we ended up at building 503 around 10:00 am and were told that the team area would not be open until noon.

We took the opportunity to open the U-Haul and examine the robots (which should have been done the night before). I was greeted with metal parts, rivet heads and robots that looked like they were beaten with a hammer. We immediately pulled BFD 3 (Big FreeWheeling Digger) out and began to go over the systems. The dumper mechanism on one side was jammed and broken. Many of traction angles on the wheels were severely bent with missing rivets. When I opened the electronics bay, drive boards poured out onto the ground. We hauled the vehicle over to the center concrete island on the old gas station and began to look for problems.


Figure 1 BFD3 Going Through Reassembly and Testing

After about forty five minutes of mechanical repair, we applied power to the system and … nothing. No response from the main ARM processor. We had no spare and were immediately on the phone with manufacturer in Arizona. My wife was checking into what it would take to catch a plane from San Jose to Arizona and have the manufacturer meet her at the airport with a new board. She would need to be back before 7:00 am on Saturday morning so we could install the board in time for impound. Denny wanted to see if he could talk to the ARM board via the serial connection but not one of the five laptops we brought had a serial port! Lucky for us, Greg’s work laptop was ancient and had the required serial port (hooray for outdated technology!). Denny was finally able to communicate with the Arm board after several boot attempts. The diagnostics he observed suggested that the SD card was having trouble being mounted. We substituted the backup SD card (thanks to Denny for making backups a couple of days before the event) and the system came up immediately. That was a huge emotional roller coaster in one hour.

Greg and Ken operated on the wheel to repair the bent traction bars (Figure 2) while Jim and I performed competition assembly on BFD 3 (hot glue on all connectors, zip tie all wires). The full functional check went fine and we shipped the robot off to impound.


Figure 2 Repairing Shipping Damage to the Wheels Before the Event


Figure 3 BFD 3 Off to Impound (Note Enormous Hangar in Back)

We unloaded our second robot (Brute Force Digger – BFD for intentional confusion) and soon word spread that we had a second excavator. We had been publishing pictures and video of BFD 3 since June 25th but nobody outside of the team had seen Brute Force.  BFD 3 was fully functional and ready to compete at the August 15th event and when the event was moved, we had more than two months extra time with very little to do. This time was enough (barely) to create a second excavator. Brute Force was created specifically to be the “hare” excavator compared to BFD 3 as the “tortoise”.

We had a parade of builders come by and look at the beast. Of all the builders, we got the most crap from Paul Ventimiglia (Paul’s Robotics). His excavator was the clear favorite and I think he was nervous about the threat Brute Force represented. The absolute top capability of BFD 3 was 8 dumps in 30 minutes at ~65 kg per dump (~520 kg total). The absolute top capability of Brute Force was 6 dumps in 30 minutes at ~120 kg per dump (720 kg total). Paul’s robot has already demonstrated 750 kg of sand in 30 minutes so Brute Force was not a threat if both excavators lived up to their potential but it was in the neighborhood.

We finished the examination and it looked like Brute Force had not taken nearly the beating that BFD 3 took. The ARM processor booted right up to a resounding cheer. Then, when the zero wheel angle command was sent, the front right wheel rotated about 30 degrees from straight and stayed there. The first words that came to my mind were “you have got to be kidding me”. The front right wheel had been having bias issues for more than a month and we had concentrated on fixing the steering axle coupling mechanism. Just before the event, we had used Loctite sleeve retainer to absolutely freeze the steering feedback potentiometer axle to the steering axle. I was dumbfounded by the misalignment. I said out loud “It is as if the pot body has become loose relative to the mount” as I touched the pot casing. Sure enough, it began to rotate.


Figure 4 Brute Force Digger at Initial Check Out

We had spare pots and one spare pot mount assembly, however, it would have taken a torch to release the Loctite sleeve retainer. The other solution was to glue the pot body in place. We did not have any 5 minute epoxy, but, we did have some 2 hour set epoxy (8 hours to cure). We aligned the pot as best as we could to the last position it was calibrated in and Jim applied the epoxy. We knew that we would not be able to go through a full functional test before the epoxy cured, but did check out most of the interfaces. Later that evening, we brought up the controller and the pot had shifted again. A quick calculation showed that in its new position, the pot would exceed its limits when we rotated the wheel to its maximum allowable turn. At that point, we broke the epoxy (which was not fully cured yet) and centered the pot to remove all biases. Jim added another layer of epoxy and we did not touch the robot again until the next morning. We had one chance for it to work right.


Figure 5 Brute Force and BFD3 at Impound

Competition Day 1

After a fairly sleepless night (what did I forget to check on the robot?) we arrived at building 503 at 6:30 am. Impound was scheduled for 7:30 am and we needed to run a full functional check on Brute Force Digger. We were stopped by event staff and told that we would not have access to the robots until 7:30 and impound would start around 8:00 am.

When we finally got access to the excavators, I checked out the steering first. The epoxy had fully set on the pot but I was still concerned that it might pull loose. I asked Jim to drown the pots in hot glue to secure them as completely as possible. We finished the complete functional checkout on Brute Force and were out of the impound area before 8:00 am. At 8:15, the event staff announced that the impound area would be closing soon. At 8:30, they announced that people really did need to finish up and get out. At 8:45, they announced that people only had five more minutes max. They finally closed the area around 9:00 am and the event could finally begin.

After the standard round of event staff introductions and acknowledgements to NASA, they were ready to pick the first excavator. Nobody wanted to go first. There is a significant disadvantage for several reasons:

1)    The event staff is still working our roles and responsibilities. There is much higher chance of a failure on their part.
2)    No one has actually gone through a real life test of the setup procedures.
3)    Since the test is broadcast to the other teams, later teams will learn more about how the regolith behaves. They will also learn how the obstacle rocks interact with the regolith.
4)    Finally, there is an absolutely huge advantage in knowing how much regolith an earlier contestant mined. If you are going for a prize, you can mine very conservatively if you only have to reach 150 kg. You can mine very aggressively if you need to reach 300 kg.

Of our two robots, BFD 3 would be least affected by going first. It was very well tested and we were confident it would easily break 150 kg (baring failures). Brute Force had much higher capability but only had about ¼ of the testing time on it. The lead judge pulled the first name out and announced it. Who was it? The first excavator was to be Team Braundo (Brute Force) of course. We quickly decided to run extremely conservatively and try to put a qualifying number on the board.


Figure 6 Carrying Brute Force into the Competition Room

Team Braundo (Brute Force)

As we brought the excavator into the starting area, we ran into an issue from being the first team. I asked the head judge when I would be allowed to install the batteries. I was promptly told that I would not be allowed to install the batteries. According to the judges, the batteries should have been installed before impound. I told the lead judge that we were told batteries were not subject to impound since we were allowed to charge them and would need to re-install them at some point. He pushed a copy of the written rules in front of me and said “Show me in here where is says you get to install your batteries”. At that point, I felt like pushing the skimpy 7 pages of written rules back into his face and going on a tirade about how many hundreds of things were not specified in the rules. Since that would not have done our team any good, I simple told him that there were many things described in the FAQ and forum that were clarifications of the rules. As far as the judges were concerned, the FAQ, forum and Matt’s emails were completely useless.

We were ushered out of the room and the judges had a one minute meeting. When they came out, we were told that since the vehicle could not be weighed without batteries, we would be allowed to install them. We were also told that we got one chance for an official weigh-in and if we were overweight, we would be disqualified. Since we never got access to the official scale before the event, I decided to install our lightest combination of batteries. Unfortunately, we had never tested with that particular battery set.

I am going to take this opportunity to make an observation. It was obvious that the rules and regulations associated with this event were never vetted by a group of experienced engineers. Nobody ever asked the question “What are we trying to accomplish?”. Does it really make sense to force teams to use a network provided by the host organization when the teams could not test communications until the day before the event? Does it make sense to disqualify an excavator for being overweight without any chance to remove hardware? Is ten minutes realistic for setting up? Is random rock placement fair?  Does it make sense to have the collector box in three different places (or tell the competitors it would be in one of three places)? This contest is supposed to be about mining Regolith, not Ethernet knowledge or who has the most accurate scale.

Recently, I have been watching the Space Elevator games and I have noted a number of differences in their competition. First, each team is allowed many runs spaced over three days. They are allowed to make adjustments between runs without penalty. They were allowed multiple opportunities to meet the requirements and when a deficiency was found, they were not immediately disqualified; they were allowed to fix it. After thousands of hours of volunteer work and tens of thousands of dollars of investment by the teams, they are being given every opportunity to show what their climbers can do. I wish I could have said the same thing about our competition.


Figure 7 Installing the Batteries

The judges weighted the vehicle and we came in at 78 kg (exactly what our shipping scale said). We removed the vehicle from the scale and the event staff mounted the collector box on back on the scale. The official timer asked us if we were ready for our ten minutes of setup. I took a deep breath and said something momentous like “OK”.

The first thing we did in setup was to drill and screw the collector camera base to the collector. This was another thing we had asked Matt permission to do so I wasn’t sure if I would get away with it. I heard a judge behind me say “Can they do that?” but I just kept going.

We got the camera mounted and the vehicle into the box within three minutes. We powered the vehicle up and waited about thirty seconds before powering our network bridge. The Ethernet bridge takes about forty seconds to connect to our router. Once that happened, we heard from Denny in the control room that he was getting data. The front right camera was not sending data (not too unusual) so Denny cycled power on all the cameras. All the cameras came back and with about five minutes left in setup, we were ready for our full functional check.


Figure 8 Brute Force in the Box Waiting for Checkout

The forward and back translation commands worked fine. The front right wheel did not respond to the turn commands and my heart nearly stopped. We continued our functional check and everything else worked properly. Although we had problems with wheel turn biases, we never had an issue with the steering failing to work before. What were the odds that the same wheel we had potentiometer problems would stop working? A problem with the pot made no sense. A loss of signal would cause the controller to become unstable, not stop working.

I knew that there were only a limited number of things I could fix in the box so I began to check those things with four minutes left. The first check was a visual inspection of the external connectors. They were all zip tied and/or hot glued so I was not surprised there was nothing wrong. I broke the seal on the front of the electronics box. The first thing I saw was the right front connector pulled loose from its drive board. Putting in the lowest weight battery configuration had pulled the connector loose (the lightweight batteries are wider than the heavier batteries). I pulled the connector back but it barely reached the board. Jim gave me a zip tie and I did my best to anchor it to the board (it was originally hot glued). Later, one of the judges told me that as I sealed the electronics back up, I said “God help us”. I think he though that was weird for an engineer to say.


Figure 9 Reseating the Right Front Connector

A final steering check showed that all wheels were working. With ninety seconds to go in our setup time, we were out of the box and ready for the rock placement. The four members in the control group (Denny, Ken, Rob and I) were isolated and the four rocks were randomly placed in the box. When everyone was ready, they gave us a countdown and then our thirty minutes were off and running.

We made our first mistake immediately when we did not click our own 30 minute countdown timer. The reason we were distracted was simple, we were a bit giddy at the rock placement. We were happy with the rock placement because it was nearly identical to one we had recently practiced on the beach (Figure 10). There were no rocks in the squares to either side of the starting cell so that opened up our preferred method of mining. Ken immediately began to execute a “slide”. A slide is where we spin all of the wheels ninety degrees and can drive sideways (Figure 11).


Figure 10 Rock Placement for Brute Force


Figure 11 Brute Force Doing the “Slide”

As part of our conservative approach, we made sure we were several feet clear of the starting box before we began to cut and stopped cutting several feet short of the box. We cut at about 60% of our normal speed and started cutting at about 50% of our available depth. By the time we finished our first cut (we can cut both going forward and backwards), I could see that the rotary cutter consumed almost the same current in packed regolith as in sand. From there on out, we cut at full depth but not at full speed.

As you can see in Figure 12, we dumped a relatively small amount of material (by Brute Force’s standard) on our first dump. Jim said that the first dump was about 70 kg. We paid a lot for our conservatism. On runs that were not at the edge of the box, there is only about 1.5 meters of linear distance to cut. By waiting to drop the cutter, we lost about 35 kg per cut.


Figure 12 Brute Force’s First Dump

Our second path began with a slide to the left. We stopped about 20 inches short of the side wall and we estimated we were only about a foot clear of the starting square. We did not want to take the chance that we were not clear of the stay-out zone so we did not drop the cutter until we were close to the middle of the box. Ken was able to push both rocks with ease right up to the wall. As we backed up, Ken steered us closer to the wall so we could keep the cutter on longer. A slide got us back to the collector and one dump later, we were on to path three. Path three was about one cutter width from path one and once again, we were very conservative on the drop and lift of the cutter.

It was about this point where we asked an event staff member how much time we had left. We were told that they could not provide that information. That put us in a bit of a bind. Of course we would keep going, but planning the timing of a last load would be difficult. Eventually, we were told that we had “about” six minutes left. We finished dumping the load we were collecting and then rushed out to get a last cut in. We hurried back to dump the small amount of additional regolith we collected. After the dump was complete and we had stopped, the official timekeeper walked in and told us we had a little less than two minutes left. Unfortunately, it takes about thee minutes minimum for brute force to collect material and return. We decided that we did not want to risk disqualification and so stopped our run at around 28 minutes. All in all, I figure that going first cost us about 50 kg due to the time uncertainty (later teams were actively kept up on the time).

After we were done, I asked the lead judge if we would be told how much had mined. He said that nobody would be told. I was very happy with that decision, it mitigated our bad draw. Other teams could not adjust their strategy based on our amount. About 20 minutes later, our total was posted for everyone to see. I asked the lead judge if I could file a protest and he told me there was absolutely no protests allowed at the event. Oh well, we were very happy to qualify with 263 kg and an excavator that worked for the entire 28 minutes.


Figure 13 Collection Buckets Dumping Into Bed (1kg/bucket – 1.5 buckets/sec)


Figure 14 Brute Force Rotary Cutter in Action

Team Summaries

I am going to publish pictures of the robots in the order that I think they competed. I will add comments on the teams I am familiar with from their web presence. If anybody wants to send me a write-up on their attempt, I will add it (and add proper credit).

Auric Designs


Figure 15 Auric Designs

This was a very interesting approach with two small lightweight excavators. It runs counter to most mining approaches (make equipment a big as you possibly can). It looked like they had unreliable communication with their vehicle.

Next Step Robotics


Figure 16 Next Step Robotics

Top Hat Robotics


Figure 17 Top Hat Robotics

I was looking forward to seeing this machine run. It would be the first excavator that attempted to use a wide penetrator to gather large amounts of regolith. Again, it looked like they had trouble with their communications.

Lunar Arc


Figure 18 Lunar Arc

Laurentian Rock


Figure 19 Laurentian Rock

This machine actually did run and did deposit some material in the box. They had enormous amounts of flotation in their tracks. It looks like they were very concerned about the rocks.

Moon Diggers A


Figure 20 Moon Diggers A

Moon Diggers A was a CMU/Astrobotics/Edison Space entry. The vehicle had an electrical short during setup. In an attempt to fix it, the team removed several parts from the Moon Diggers B robot which was still in impound. As a result both Moon Diggers A and Moon Diggers B were disqualified.

Technology Ranch


Figure 21 Technology Ranch

I expected Technology Ranch to do very well. They had been in the competition since the beginning and almost won it the first year. They were certainly capable of putting up a 200+ kg number under the right conditions.

Unfortunately, they were unable to get their cameras working. They still managed to deposit 26 kg of regolith using blind dead reckoning! Officially, they were disqualified for digging in the starting cell. This reinforced our decision to stay well clear of the starting cell.

C2 Robotics (SLO)


Figure 22 C2 Robotics

The scoops on this monster were enormous and driven by a 4.5 HP BattleBots motor! I expected them to move an impressive amount of material. Their wood guide was a very smart way to make sure they were lined up to the box for a dump. Unfortunately, they were unable to get their cameras working (do you sense a theme here?) and drove over a rock at the start.

Colorado School of Mines


Figure 23 Colorado School of Mines

Innovation Island


Figure 24 Innovation Island

UBC TREAD Robotics


Figure 25 UBC TREAD Robotics

Team Waldbaum


Figure 26 Team Waldbaum

Since the first day videos of this excavator appeared on the net, it stirred up a lot of talk. Most people thought it fell under the category of “barely mobile excavator”. Over time, I thought that the design changed to be more in line with the conventional excavators. With so much weight and power devoted to cutting, it had the potential to cut enormous amount of material. Its ability to dig very deep and drive over rocks (if needed) gave it potential for first or second place. Unfortunately, it had network issues and after almost two hours of working with the event network crew, it didn’t mine a single gram.

Paul’s Robotics


Figure 27 WPI Robot at Impound

It was about this point in the competition that I knew there were not many more excavators likely to break Brute Force’s number. The entry from WPI was not only expected to break the number, it could easily double it. Coupled with their powerful excavator, the WPI team got a near perfect draw in the competition.  They were the 14th robot (out of 19) and got to see every major competitor go before them except BFD3. They could have run at 50% capacity and still won easily.

I have known Paul for a number of years and we both have competed in BattleBots. Our teams were in constant contact while we were building and testing our robots. I knew all of the time and effort that went into the design, build and testing of their robot and I had no reservations about them being the one to beat us out for first prize. Denny told me that their network people knew about the “secret sauce” for communicating over a delay and they were obviously well rehearsed and well prepared. Still, NASA awards money for performance, not potential.

I told my wife that there was a 99% chance Paul would beat our number and I was fine with that. I also said that there was a 1% chance something would go wrong. They might be over weight, something might break, there might be a driving error or they might do something to disqualify themselves. In any case, I told her that if that happened, I would not feel guilty about it because our team had put in 4000+ hours to earn a 1% chance at first place.

When I first saw their robot in action on the web, I noted one major design flaw which I pointed out to Denny. Ironically, in the end it was not only the many strengths of the robot but the one design flaw and one other weakness that allowed NASA to declare Paul’s Robotics the winner. More on that later.

As expected, the team had no trouble (that we could see) in setup or with the network. The judges placed the rocks and with the first movement of the robot, their victory was practically guaranteed.

Right out of the gate I noted that the treads were having a very difficult time with traction on the packed regolith under the fluff. During their higher speed maneuvers, it looked like they were only getting about 50% vehicle motion relative to track motion. There was far too much floatation in the tracks and when they were empty, there was not enough pressure to penetrate the regolith. They had adapted their robot to work in soft sand and it was not performing as well in regolith. Still, it was relatively minor and I expected Paul to quickly adapt. I did note that their first cut was very close to the stay-out zone but after that cut, we had trouble seeing the regolith surface due to the low lying fog of dust .


Figure 28 WPI Dump After First Cut

There was a point in the run where several of the cutting channels got damaged when they tried to penetrate too deep in the packed regolith. The damaged channels threw the mined material up into the air when they turned the corner at the top of the excavator and made a distressing sound. Despite the damage, the cutter held together for the entire time. Although the vehicle got stuck once, it was clear that it had deposited significantly more regolith than Brute Force. I knew that there were a few more capable excavators to come, but our excavator had a very good chance at third and an outside chance at second.

It was at this time that things began to happen that I now wish I had never seen. The judges appeared to ask Paul to move his excavator to the far side of the box. The audience was still looking from the collector side camera (about the POV of Figure 29) and the judges began doing things in the box. They started by drawing the stay-out zone in the regolith surface. From our vantage, it was difficult to see what held their attention. I was hoping that they were not trying to interpret some vague marks as mining in the stay-out zone (that was our biggest fear of this rule). I went out to Paul (who was waiting with his team between the building and the tent) and told him that the judges appeared to be checking whether he mined in the stay-out area. I told him that he should insist that he get to review their findings. The white arrows in Figure 29 shows the marks they were making in the regolith to outline the stay-out box.


Figure 29 Judges Marking the Starting Square

When I got back in from talking to Paul, the A/V engineer had changed to the reverse angle camera (about 180 degrees from the first). The crowd was buzzing with conversation because we could see an enormous trench cut very close to the stay-out zone. The trench was actually the right side of one of the excavator cuts. The edge was pristine, straight as an arrow and about 1.5-2.5 inches deep. Figure 30 below shows about where the cut started and ended (marked in blue).


Figure 30 Location of the Right Side of Cut (Blue)


Figure 31 Video Details of Cut In Question


Figure 32 Picture with Starting Box and Cut Marked

One of the judges came into the box and carefully studied the cut. He then went over to the excavator and measured the width of the cutter with a measuring tape. He went back to the beginning of the cut and laid down his measuring tape perpendicular to the right cut face and there was a ridge where the cutter would have dropped. You can see the ridge in the regolith in Figure 29 (marked with a blue arrow).  It is clearly in the stay-out box. After he left, a different judge came in and repeated the same process. His perpendicular tape location was also clearly well within the stay-out zone.  Over the next few minutes, judges came in both singly and in pairs. We could not hear them but in no instance did one judge appear to be disagreeing with another. In one case, a judge made a scooping motion with his hand over the corner of the stay-out box. He was clearly indicating that material had been removed from the corner. Figure 33 is a blowup of Figure 29 with a line added. It is obvious that significant material is missing from the corner of the stay-out box. In the blow-up, you can clearly see the ridge where the cut starts. It matches where the slope changes on the front line.


Figure 33 Material Dip in the Right Corner and Cut Starting Ridge

I certainly felt sorry for Paul. He clearly had the most capable digger at the competition. After all of his hard work and testing he was going to leave without any money. It was a crappy rule and one that served no purpose.

On the other hand, the stay-out zone rule had already disqualified Technology Ranch. Paul was fully aware of the rule and could have chosen to be extra conservative as we were when we drove Brute Force. Instead, he relied on canned routines to get him out of the stay-out zone. He did not design his tracks with compacted regolith in mind and the slippage caused from the excess flotation caused him to drop his cutter in the stay-out box.

What finally made me certain that he was going to be disqualified was arrival of a judge that came into the box and commenced taking pictures of the area with the measuring tape showing where the cutter was when the trench was cut. They would certainly need to carefully document why they disqualified the best digger.

Out in the viewing area, the rest of the teams were actively discussing what was going on. Somebody suggested that the builders sign a petition to exempt WPI from the starting cell rule since they had mined so much material. The vast majority of teams were against such a petition. Like me, they thought that even crappy rules should be applied without bias. An excavator had already been disqualified for mining in the starting cell. It should make no difference how much the excavator mined.

The judges went into conference for more than twenty minutes. We could see them standing in a circle and the discussion appeared to be very animated. After they finished, the head judge came to the podium and declared that Paul’s Robotics was “qualified”. He left immediately without taking any questions. I guess that somebody told him the camera was on the whole time they were examining the regolith box because he came back to clarify. He said that the judges could not come to a conclusion on whether Paul’s robot mined in the starting cell or not. A few minutes after that, a staff member came out and confiscated the tape recording from the A/V center in the tent.

There were a lot of shocked people in the room. I believe that all of the judges are engineers and I cannot fathom how they could reason:

1)    We know exactly where the right side of the cutter was
2)    We know the cutter was down about 2 inches
3)    We know that cutter was traveling in a straight line
4)    We know exactly how wide the cutter is
5)    We have a ridge at the start of the cut, the ridge is within the stay-out zone
6)    We know that the cutter must have been down in the stay-out zone
7)    The evidence is inconclusive

Did they believe that the vehicle was levitating on the left side when it was cutting on the right side? Where did the ridge come from? If there was any question about whether the cutter was cutting over its full length, they could have reviewed the video tapes of the run. Although they could not see the surface of the regolith due to the dust, they could have verified the angle of the vehicle, the linear motion, the material being ejected from the cutting blades. The tapes were immediately available and with $500,000 on the line, every effort should have been made to get the decision right.

My take on the whole situation comes down to this. The judges were looking for any way to avoid disqualifying the vehicle that just mined a huge amount of regolith. At the time, there was only one “qualified” vehicle and disqualifying Paul’s excavator would look really bad for the event. If the evidence was monumentally obvious, they would have had no choice. Once they found some wiggle room, they wiggled out.

In no way does this action reflect on Paul. The judges came to their decision without input from him. Even it he knew that he had mined in the starting cell, would you expect him to ask to be disqualified? I think not.

This brings us full circle to my assertion that two negative characteristics of the robot contributed to Paul’s Robotics winning the event. The first characteristic was the robot’s prolific generation of dust during mining. The excavator generated so much dust that they had to pull the judges out of the competition box area. During all of the other team’s runs, the judges very carefully observed and marked the starting locations of the cuts as the excavators left the starting zone. Since the judges were removed from the box area and the dust cloud was virtually impenetrable down low, they had no chance to document the start of the cut in question while it was still pristine. The second characteristic was a design fault that I observed on first seeing the vehicle. I noted that the low bars between the tracks could result in the excavator being “high sided” if a track dropped into moderately deep cut. That is exactly what happened during Paul’s trial. The vehicle got stuck on the left side of the illegal cut while trying to return to the collection box to dump. Paul spent a minute or two moving back and forth to free the excavator from the ridge. In the process of moving back and forth, he removed most of the traces of the left ridge. Without the left side to make the disqualification completely irrefutable, the judge found an escape hatch.

For me, the worst part is that I now feel very unhappy about the competition. If Paul had run a clean run, I would have been ecstatic about getting both second and third. Now, I just feel betrayed. If I could find a copy of the video from Paul’s run and the judges actions after, I have no doubt that I could prevail in court. But at what cost? Even if I won, I would probably only gain about $200k after legal fees and Paul might be out $500k. The only win there would be for the lawyers. I am hoping that by documenting it, I will get it out of my system and move on. One final note on this. I am not the only one being “screwed” by this. If Paul has been disqualified, the unclaimed money would likely mean that the event would be run again and the other excavators would get another chance at that money.

Invading Huns


Figure 34 Invading Huns

I was looking forward to seeing how much this vehicle could scoop in a single pass. When I heard that they were running autonomously, I wasn’t sure how their system would be able to satisfy the traversal rule and dig multiple times. Their single scoop of material was extremely impressive, however, it never made it into the collector. They claimed to have made a last minute code change so that they wouldn’t back up into the collector with as much force. This chance was untested and is what they believe lead them to miss the dump.

Sandstorm +


Figure 35 Sandstorm+

We were really rooting for this one-man team. He was right next to us in the robot pit and we tried to help he were we could. I believe that his vehicle could have mined a significant amount of material, but once again, the network won.



Figure 36 E-REX

Another one man team. This vehicle did fairly well and had one very large dump. He officially placed 4th. It seemed to have trouble with commands and/or communications. He was using a stock web-based utility for his webcam. He had to setup the initial connection via TCP, but once that occurred the images went over UDP. Another thing to note, this vehicle clearly cut a swath right through the starting cell, but nobody said a word about it.

Green Cheese Solutions


Figure 37 Green Cheese Solutions

Big FreeWheeling Digger 3


Figure 38 Transporting BFD 3 to the Box

So it looked like we were going to book-end the competition. Brute Force started first and BFD 3 was the last name pulled out of the bag (342:1 odds against that). This excavator had probably been tested longer than any other excavator in the building. I was not particularly worried about whether it would start or run.

The team had decided that we were not going to try and go after first place. At maximum capacity, the hopper only held 65-70 kg of regolith. We would need 9 or 10 dump to get close. The most dumps we ever accomplished on sand was 8 (although the movement speed is only about 50% on sand compared to what we expected on regolith). We would need to run the system about 50% faster then nominal for both cutting and dumping. We had the power and force, but, Denny did not feel comfortable driving at that speed. You can waste a lot of time escaping from a driving error.

Since we were not going for first, we wanted to break 150 kg without being disqualified. Like Brute Force, we dropped the cutter late and pulled it up early. Unlike Brute Force, I knew I could burry the cutter to max depth immediately.

Setup was uneventful. We finished with 3 minutes left and I had time to take a leisurely look around the excavator. Our biggest distraction in the control room was freshly delivered KFC chicken and all the trimmings. After all of the failures in the event, the NASA staff seemed to recognize and appreciate how professional our controllers were (thanks Aaron and Denny!). As you can see in Figure 39, the panel includes the live video feeds, vehicle telemetry in text form, vehicle telemetry in graph form and a clear set of instructions on how to run the vehicle.


Figure 39 Excavator Control Interface Extraordinaire!

After the rocks were set, we entered the control room and remembered to set our countdown timers. The rock placement was an absolute joke. It was like somebody watched Brute Force’s ability to slide and placed rocks in the four corners to prevent it. BFD 3 does not slide (physically, it can but the wheelbase would be very narrow and it might tip over). BFD 3 works best when it carves circles and with the rocks in the corners, we didn’t need to move a single one.


Figure 40 BFD 3 CCW Circular Cut

There was only one scary moment in the run. When we installed the lexan box that held the router, it was put too far back on the Velcro strips. The first time that the cutter was lowered all the way, the teeth on the buckets began to catch on the tape that held the lexan box together. Every time a cutter came by, it yanked the box up and then dropped it back down. Eventually, the box was pushed farther forward and the teeth stopped hitting it.

Of the two vehicles we fielded, I am most proud of how the team created BFD 3. This machine was designed and built to be a regolith miner, not to win a 30 minute contest. In comparison to all other vehicles at the event, operations are smooth and continuous. If the contest called for mining the most regolith over a month, I have no doubt that BFD 3 would win by a mile. Round and round it went, clockwise cuts followed by counter clockwise cuts. No effort wasted, no strain. The vehicle still consumes less than 150 watts (last year’s requirement). When the vehicle is not performing a fast wheel turn, it consumes between 60 and 90 watts. Think about that the next time you change a 100 watt light bulb.

In the final tally, we got 7 dumps in 30 minutes. Our super-conservative approach netted us 270 kg of regolith. Good for second place.


Figure 41 BFD 3 CW Circular Cut


Figure 42 BFD 3 Dump

Moon Diggers B


Figure 43 Moon Diggers B

After the last vehicle name was pulled from the bag, the lead judge announced that the Moon Diggers B team was “un-disqualified”.  The official reasons were never given.

The vehicle is one of the few that actually moved in the box. It could have moved a fair amount of material, but the techniques they used were based on sand. Much of the regolith they gathered fell out of their scoop as they were trying to scoop more. When the scoop was set too deep, the vehicle tended to dig itself in. Eventually, the front right wheel stopped working.

2009 Regolith Excavation Summary

We came, we saw, we almost won. We got enough money that I have a chance of convincing my wife that not all of my ideas are without merit. After two years, the team is pretty exhausted. I am incredibly lucky to have friends with amazing skills to draw upon.

What next? In the short term, the team has been invited by NASA to participate in a in-situ resource utilization exercise this February. Fortunately, it is in Hawaii so the team can take some much needed rest after their work is complete. In the long term, I have been hoping to enter the NASA power beaming challenge with an approach unlike any fielded so far. Of course, they have not announced if they are going to run it again.

That is all for now. Keep thinking outside the box and never give up!


back home

October 20, 2009 - 8:39 am 1 Comment

the team and robots arrived safely home. in case you haven’t heard from the news media, bfd3 got second place, and bfd4 got third place. a full write-up will come as soon as the numbness wears off.

bfd4 qualified

October 17, 2009 - 1:31 pm Comments Off

the bfd4 was the first robot to compete in the challenge. to be honest we expected it. i guess we desrved it for being so secretive about the vehicle.  bfd4 excavated 263.75 kilograms.. we had a slight panic when we powered up.. the front right steering motor became disconnected despite being hot glued in. todd was able rig something up with a ziptie. as of this posting 2 other vehicles were not able to collect any material.

challenge eve

October 16, 2009 - 8:58 pm 1 Comment

The cat is out of the bag. Yes there is two vehicles. I will attempt to upload photos/videos from the competition of both vehicles. for reference, they are referred to as bfd3 (big freewheeling digger) and bfd4 (brute force digger).  bfd4 has a completely different cutting mechanism compared to bfd3. you might notice that the cutter design is based upon our entry from the first competition.

As you might have read in the previous post, transporting the vehicles turned out to be a nightmare. However, I am happy to report that both vehicles are doing fine. we did panic a little bit about the damage that bfd3 sustained. we spent a couple of hours fixing it and putting it into final configuration for the competition. by the end of the day, we were able to test with the competition wanem machine and verify that all systems work great. what a relief.

It was really great to meet all the teams that showed up early. Most people are in this competition just for the fun of it, and the general sportsmanship has been really nice. It’s hard to understand what most of the teams went through to compete in this competition. I think everyone is numb from all the hard work that we have all done to prepare for these next two days.

see you all tomorrow!

we arrived

October 16, 2009 - 1:27 pm Comments Off

well we made it up, but the bfd sustained some serious damage on the trip. it appears that one of the straps came loose and the bot shifted around and slammed into the back door of the uhaul. the right dumper door spring was ripped out and the bolt was sheered. several treads were severely bent and one of the wheels was bent. inside of the electronics box, 3 of the motor drives managed to break off. also the sd card holding the software managed to have failed. the system fails to boot.

we have a ton of tools and spares, hopefully we will manage to get the vehicle in an operational state… we will try to update you guys later.


October 14, 2009 - 6:21 pm Comments Off

well.. the packing as already begun. final checkout is complete. next stop nasa ames.

team listing

October 7, 2009 - 9:26 pm Comments Off

Where does one go to get the latest official regolith excavation challenge news? Nobody really knows. With an official regolith website, a forum, a twitter, and a facebook, you think you could check all of those every day and get the latest news. Nope.

The latest news is now on the csewi webpage! Clearly the people running the competition are having social networking anxiety by making it increasing difficult to access information.


Anyways, here is the official team roster:

we destroyed our cutter motor

October 4, 2009 - 5:43 pm 3 Comments

As the competition gets closer, the probability of a major vehicle failure increases exponentially. Right on schedule, we managed to destroy a pricey cutter motor 2 weeks before the competition. Unfortunately, the motor has an 8 week lead time. The motor highlighted in this story is used to drive the chain that moves buckets down into the regolith and back up into the vehicles collection bin.

As you can see below, the drive shaft is slightly bent. It doesn’t look like much, however, it causes about a 3 amp variation in power consumption when running with no load.

How did this happen you ask? Well, since we do not have a fullscale regolith box to test in, we tried to test our excavator in some very hard packed dirt. The only problem was that, we didnt know there were rocks embedded in the dirt. The BFD dug into one and stalled the motor. As you can see below, the motor jumped up to almost 12 amps! We drew less that 4 amps cutting in that hard dirt before we hit the rock, which is pretty good.


LCROSS Activity Day

October 2, 2009 - 9:32 pm 1 Comment

If you are reading this site, you are probably interested in stuff related to the moon. If you are interested in stuff related to the moon, then you are super excited like us for the LCROSS impact event next friday morning (8/9/2009). If you haven’t heard of it, then you go better go read this:

As some of you may know, Northrop Grumman was responsible for building the LCROSS spacecraft. All of the TerraEngineering team members work at Northrop Grumman. In fact, when LCROSS had that glitch in the IRU, Todd was one of the engineers that was part of the anomaly resolution.  (Needless to say the spacecraft is working great now).

Anyways, Northrop Grumman is having an LCROSS Activity Day (10/8/2009) on the main Space Park campus. They have asked us to demonstrate the BFD. We plan on bringing the arena and doing some mobility demonstrations on the pavement.  There should be a good turn out!