Saturday, October 25, 2008

ETH Competing Tonight

We will be starting tonight, as third team out of four. With the experience of last night, it’s really hard to predict, when this will be. According to the schedule we’re supposed to start at 1:30 CET. But this can be delayed or changed without any further notice. So if you’re interested, it’s probably best to check the ESA life stream once in a while.

The stream can be found here:
rtsp://195.169.140.121/Live1.sdp
(or if you are behind a draconian firewall rtsp://195.169.140.121:80/Live1.sdp)
rtsp://195.169.140.121/Live2.sdp
(or if you are behind a draconian firewall rtsp://195.169.140.121:80/Live2.sdp)
One will follow the robots, one will cover the controlling team

Lunar Robotics Challenge - Day 5 – First Day of Actual Challenge

Our team spent the day fixing hardware. In the morning, we decided that there is not much use of additional testing. We will not have the time to change and improve things anyway, but it would introduce a certain risk to severely damage the robots and not being able to compete at all. Instead, we used the time to improve our hardware - especially making it more robust. The pajamas have been working fine so far, but as the robot has to travel long ways on really rough terrain, we thought it might be better to enforce them. Also we exchanged almost all cables of the pully-system, hoping to prevent bad surprises like on Thursday.
At 9 pm the challenge started with the team from Finland, who where followed by the Jacobs University Bremen and Madrid. The team from the UK was unfortunately unable to start due to three
The base camp at night. It’s a really magical place at night. It’s buzzing with students that work in the dark outside their trucks, giving the last finish to their equipment before getting to start.


This is how the way to the actual crater looks like.


The team of the Jacobs University Bremen are about to start. While the Lander is illuminated with this very weak side light, the inside of the crater is completely dark.

Lunar Robotics Challenge - Day 4 – Day and Night Testing

To be honest, there wasn’t a lot of testing for us today. Just after we carried all our stuff to the top of the testing hill, the WiFi-adapter on the walker broke down. After several unsuccessful attempts to fix the problem via software, we had to take the whole robot apart and actually install a replacement card. Then we could do a couple of steps downhill that were really promising with regard to the stability of the robot on the slope. Even without being supported by the tether cable, the robot didn’t slip an inch. The scale-like plates on the bottom are doing a great job. Unfortunately after getting to deep into the crater, we lost communication and had to abort the trial. And at this point everything started to go wrong. First the hardware broke down. We had to replace some of the cables in the pulley system, exchange one of the bevel-gears, and keep patching the pajamas. At the same time, we had to work on the communication issues until deep in the night, so that we couldn’t do any of the testing we were supposed to do (like spotting the specimen, collecting it, and generally navigating in the dark). It was a bit frustrating after the extremely promising first trials on Tuesday and Wednesday. But then again, it’s probably better to have to the trouble now, than during the actual challenge.

The two robots right at the edge where we were testing today. The slope is not as steep as it appears in this picture, but even for humans, it’s really hard to get up the sandy inclines.








Our team on the top of our testing peak. It’s Tenerife, but we’re at more than 2000 meter altitude. So it’s not too warm. From left to right: Andi Lauber, Cédric Pradalier, Martin Latta, Oliver Baur, and David Remy.

Thursday, October 23, 2008

Lunar Robotics Challenge - Day 3 – Rain and Other Troubles

The Tenerife-Virus hit the camp. Eight people got hit by this nasty sickness that takes you out completely for 24 hours. Unfortunately, Cedric and I were among the lucky winners. So no Lunar Robotics Challenge for us today.
Martin, Andy, and Oli finished the sealing work on the walking robot in the morning, and adapted our gaits and calibration to the new configuration (with scales and ground plates). Unfortunately, once they were ready to test, it started to rain at the camp. So we have to do our big all systems test tomorrow. On the other side, they were able to take advantage of the time, and implemented additional motions for the walking robot. We’re now able to crawl way faster. Traveling about two meters/minute (which should now be fast enough to complete the mission in the given time span of two hours). We also implemented some motions for fine adjustment and collection of the soil samples. Tomorrow, ESA will provide the actual sample material, so we’ll know how this will work out.

Lunar Robotics Challenge - Day 2 – Getting our Robots Ready to Run

Today was the day we gave our robots the final touch and had them playing in the sand for the first time.
Andy and Martin spend the entire day working on the walking robot. For the first time, we put on the pajamas (which miraculously fit) and made everything sand proof. Also the metal ‘scales’ on the shanks and the main body were mounted. We are really happy to see that the range of motion was not reduced by these additions. In the late afternoon, we were able to make some first crawling trials and saw that the robot behaves very nicely in sand. The motions are smoother then on the hard floors of our ETH-lab, and it seems that the robot has the climbing abilities that we hoped for.

Cedric ran some tests with the Crabli, which was being operated in an outside environment for the first time, as well. It worked as we hoped, no mayor problems with the steering and communication. The only issue we have to deal with are little rocks, that can get stuck in the wheels of the rover and block them. So I guess, there will be some tape, paper, and maybe tooth brushes to form some simple protection during the actual challenge. I was working on the SpaghettiBOT (as we started calling our winch), but (besides tons of Blue-Screens) nothing really special happened there.

Also the starting order was determined today. We will start as team 7 of 8. On the upside that means that we have one more day of testing then the others and can learn from them. On the downside we have to start around 1am (local time) Sunday morning when it will be really, really cold! If you want to watch this event, ESA will provide two live video streams during the actual challenge. They will be visible with Quicktime or VLC and can be found at the URLs:
rtsp://195.169.140.121/Live1.sdp
(or if you are behind a draconian firewall rtsp://195.169.140.121:80/Live1.sdp)
rtsp://195.169.140.121/Live2.sdp
(or if you are behind a draconian firewall rtsp://195.169.140.121:80/Live2.sdp)
One will follow the robots, one will cover the controlling team (freezing in the night). Our turn is Sunday morning at 2am (Zurich time), so this might be a nice thing to watch after coming home from going out.

Lunar Robotics Challenge - Day 1 - Setting up Camp

We arrived at our hotel very late on Sunday night. It’s a really remote spot almost on top of the Teide volcano on Tenerife. After the 12 hour trip, we were joking that ESA set up the event here, because it’s almost as remote as the moon itself. Then again, the scenery really rewards for the long trip. We had to climb more then 2000 meters, starting right at the coast, through a rain forest, until we reached a desert-like landscape in the bottom of an old crater. It’s a truly lunar scenery with petrified lava-streams and weird rock formations that give the whole setting an extremely extraterrestrial touch. Especially when driving through it at night.
Our first day started rather slowly. ESA was putting up the challenge site on which we took a short stroll before we explored the area around el Teide. It was really nice to be at the actually testing grounds and to see what our robot has to handle. During the last months, the only information we got about the challenge came from the blogs of our competitors and google-earth. It gave us a rather abstract and competitive feeling about this event. But seeing the crater today, we realized that the challenge is not at all about being better then the other teams, but about completing the mission scenario. But even if we fail to do so, the entire project will be a great success. Just getting a robot here and being able to test it in an environment that no lab in the world can provide, is more then we could have asked for. The crater seems bigger in reality then we imagined it. It’s also way steeper then we thought and the bottom is covered with big rocks. The soil is made up of granular volcanic rocks (with a couple of millimeters in diameter) that are super-light and, despite their size, almost behave like sand. So we’re really happy to have a tethered solution that will help us get the robot uphill.
On the picture above, Andy is marking the lander’s location in the far right background. Martin is standing on the edge of the crater and David where we expect to get our soil samples. Considering that our robots are less then half a meter long, the distances are considerably.

Our urges to start immediately were hold back by a lack of robots, which arrived at around 4 pm at the testing site. Thank god, nothing was damaged during shipping. Once they were there, we instantly set up shop in our truck before we had to go to the security debriefing at our hotel.

Wednesday, October 8, 2008

video

This was the first complete system test, less than 20 hours before the departure of the robots. In this test we had:
  • The ASL Walking robot (complete with cover and lights)
  • The ASL Crabli (6 wheeled rover)
  • The ASL SpaghettiBot (Power cable manager)
The test was done in the dark, with the three controllers (David, Oli and Cedric) only using the robots cameras for their situational awareness. The goal was to "find a white object" and they had no prior knowledge of the room configuration. All communications from the landers (SpaghettiBot) to the Crabli and Walking Robot where done through a multi-hop wireless network.

The video is a recording of what was visible via the Crabli's camera, sped up 5 times. This camera was also used to drive the crabli, not to record the experiment. As a results, the image has saccades and sometimes focus on weird things such as the Crabli's own wheels.

Even if we would have used the opportunity of more testing, we are quite happy that our infrastructure is now working properly. Our control stations are stable and usable, our wireless network can handle the load, our night perception is acceptable and our locomotion capabilities have been proven.

How will we do in the sand outdoor? Let's wait and see...

Done & Done


Final Tests Before Packing







Here are just some pictures from our last night with the robot ensemble. It was a lot of fun, but at the same time, we realized how hard it would get to navigate with the robots vision alone.

Migration to the new Housing Finished


Today we finished the migration of our hardware into the new frame and housing. After fixing the usual problems with the off-the-shelf components and some teething troubles with the wiring, we could run our first successful ‘walking’ tests just this weekend. The housing now contains the complete robotic system, including cameras, illumination, communication, and –most importantly- the specimen collector.
In contrast to our initial plan of having a drill-like collector, we decided to go for a simpler shovel mechanism that is just lowered into the ground and filled by slowly sliding the robot forward. It looks a little bit like a bird’s beak.
Mechanical benefits are the increased stiffness with some reduction in weight, and additional sensors to measure all joint angles. Additionally the robot is now fully protected against environmental influences. However, we’re not planning of putting it in the rain until the competition is over.
The upcoming week is fully designated to testing. The main focus will thereby be on the cooperation of the individual systems. That means, looking at the communication, power consumption, and maneuverability in the dark. Another goal is to get some long term operation experience before we have to pack our boxes and ship them to Tenerife. We want to figure out witch components are working well, and which are prone to error while we’re still able to patch. Other issues that we couldn’t yet examine are friction, wear, and especially heat accumulation and its influence on the different components.
The fine-tuning and finalization of the locomotion concept will be done on the spot in Tenerife were we can test in the actual conditions and environment of the competition.

Saturday, October 4, 2008

Rollout of the 1.2-Protoype at the Night of Science in Zurich


Everything was on an extremely tight schedule for the preparation of the Night of Science. After pulling some long, long evenings, we finished the new prototype literally in the very last minute. Some of the final assemblies were actually done in the exhibition tent. The great effort of the students was rewarded by an extremely open and interested crowd that was drawn in large numbers to our booth.
The only drawback was a small error in the NI-software. It sometimes caused misinterpretations of the sensor signals and made demonstrations impossible. Still, we got good feedback and the new prototype was a real highlight in the exhibition; especially for the younger visitors.

Thursday, October 2, 2008

Spill Winch for the Tether System


As we decided to have a tethered power supply (and climbing aid) for our legged robot, we were in need for some device to pull the power line and the robot out of the crater. After unsuccessfully looking for a traditional of the shelf solution, we decided to create our own spill winch. This design has the advantage that there is no need for slip ring connections and that the effective cable pull is easy to control by the small pre-loading system.

Wednesday, October 1, 2008

Supporting Robot ‘Crabli’ is Test-Ready


After Thomas finished his test series on different suspension systems on the ‘Crabli’-Robot, we were able to mount our additional equipment on it. Besides an extension battery pack, we added a (night vision) camera, a wireless repeater, and a strong LED light source. Mark is currently simplifying the control interface and we’re running tests focusing on the remote controllability and the range of the WiFi signal.
For this robot the LRC will be the first field mission, and we’re really curious about its locomotion abilities in such a harsh environment.

Monday, September 1, 2008

Development of the GUI and Communication started

Slowly we have to deal with the question of how to control our robot from afar. The computational functions like generation of joint trajectories and their realization in the controller have to be equipped with well defined interfaces and have to be able to communicate with each other. While this has been mostly solved for internal communication (using the TCP-IP connection), we are now working on the communication with the operator. Dealing with questions how the motion of the robot is being provided and controlled.

New Computational System Successfully tested


In a breadboard environment we evaluated the idea of directly reading the sensor information via the cRIO system. This would eliminate the need for additional servo boards at each leg and frees the CAN-Bus from some of its load. We also would have additional 16 analog channels available, which can for example be used to monitor the motor temperatures. Digital filters were implemented in the FPGA and the control code was ported to NI LabView were it runs in real-time at with a higher frequency.
So far the move seems to be a great success. And with the new singleboardRIO, it will become feasible to have the computation on board without adding a lot of additional weight.

New Backbone Ready




The workshop has finished manufacturing of the new robot frame. Currently the composite casing is being built. Together they will provide a very stiff framework for the mechanical components and will also provide a dust-prove housing for the electronic components. With this step the mechanical redesign for the robot is finished and we can move on to integrate the specimen collecting mechanism. We also created a prototype for the textile cover of the legs. This was one of the critical issues, as we suspected that part of the fabric might jam the joint motion. However, preliminary test seemed promising.

Friday, July 18, 2008

The robot did his first walking moves controlled through the CompactRio-Device of National Instruments. The next step is to implement the actuators. Furthermore the robots movement is now only specified by position and we want to add the parameter velocity as well. So I still got some things to do...

Monday, July 14, 2008

Concepts presented @ CDR



Our basic concept is a team-based implementation of two tethered devices: a wheeled rover providing overview sight, power supply and a communication bridge and a legged climber that will fullfill the main tasks as descend into the crater, sample some soil specimen, climb up the slope and return the samples to the landing site.

A fully covered second prototype of the legged robot ALF will be constructed and manufactured soon. The higher torsional stiffness allows more accurate control for reliable walking abilities and the cover protects the interior against sand and dust.

As a sampling mechanism, we are suggesting a drilling device (Archimedes screw) to draw the granular specimen. It will be located at the front and the suction procedure will happen in an animal-like way.

CDR @ ESTEC

From the 9th to 10th of July, two of our members (David and Andi) went to ESTEC in Noordwjik to participate in the Critical Design Review of the LRC. Many impressions about the LRC, the other Teams, their ideas and ESA itself turned the short trip to an exciting event. We're looking forward with enthusiasm to the TRR and the competition in fall on Tenerife.

Sunday, July 6, 2008

Visit Einar Nilsen

Last Friday, Einar Nilsen came to visit our project. He works at the school of applied science in Buchs and has ample experience in the design of robotic systems. We showed him the existing prototype and discussed the drawings and sketches for our new version. It was great to get his input on these issues and we’re hoping to continue a collaboration with him.

The Start

With the advantage of having an existing prototype at hand, our first task was the identification of its strength and weaknesses. We looked carefully at the mechanical design, the electronics, and the controller, thereby defining the alterations we want to make to the existing system.

Mechanical Design:
The design of the legs proved to be well done. They are robust, flexible, and still lightweight and compact. We will have to make some small adaptations to the sensor placement and the integration of the elasticities. The only thing that will be completely new are the feet. They will be specially designed for a sandy environment and equipped with some climbing aids. But apart from that the overall leg design will remain the same.

In contrast to that, the main body will undergo a complete redesign. We need to increase the torsional stiffness, make room for the specimen collector (which also has to be developed), and increase the robustness such that the robot will survive a fall or tipping over. Last but not least, we have to cover everything and protect the interior from sand.


Electronics:
Adapting electronics will mean mostly adding components: Camera systems, wireless communication, additional sensors, on-board control, and power supply. We will use the same actuators as in the existing prototype as they turned out to have a great balance of size, weight, and power. We will add simple footswitches to detect ground contact while the robot is upright (we first wanted to actually measure the forces, but considering that the robot will undergo various leg configurations, this seems just impossible), there will be inertial measurement, and some supervising sensors. And most importantly, we will focus on a tight integration of electronic components into the mechanical hardware.


Control:
The development of the controls is naturally one of the most important parts in such a project. Right now we’re working with predefined trajectories. I.e., a set of open loop joint angles that lead to walking, turning, etc. This collection will be extended to include tasks like standing up, or turning back into an upright position, crawling, crabbing, and the like. The long term goal is, of course, the generation of the motor inputs on line, to allow flexible navigation and at some point fast and dynamic locomotion.

Friday, July 4, 2008

The Roots

As you can see below, our project group is a true student team. Andi, Martin, and Oliver constitute the core. They all just finished their Bachelor’s degree and agreed to work full time on the LRC project until they will continue with their Master’s. They will get additional support from a number of PhDs from our lab, and other undergrad and graduate students who, over some duration of time, will extend the team and contribute to specific subtasks.
Our LRC platform is based on the undergrad design project of Andi, Martin, and Atilla Yilmaz. During the last semester, they were developing a quadrupedal robot with series elastic actuation on the main joints of the legs. The goal was to develop a walking machine that is able to passively adapt to irregular terrain while walking statically, and can additionally use its elasticities to store energy during dynamic locomotion. Without knowing of the Lunar Robotics Challenge, they had named the robot ALF, already foreseeing that its future will not be limited to earth.
The project was a great success and we are certain, that a similar robot will perform well in the Lunar Robotics challenge. However, it is evident that we’re not even half way there yet. Major adaptations, improvements, and extensions have to be made to the original design, to get ALF space ready. This task will occupy us over the next couple of months.




Wednesday, July 2, 2008

Fabian (Teammember)


My name is Fabian Seitz and my part is the implementation of communication tasks and assisting Oliver in software programming. This means to develop a framework in order to initialize and control the robot comfortable.
I'm studying mechanical engineering at ETH and focusing in my graduate program on robotics and autonomous systems.



Oliver (Teammember)




Hi there! My name is Oliver Baur and my job is connecting Martins and Andis work. Which means programming the software so that the trajetories of the stable walking gaits actually move the legs of our robot in the manner we want them to move.
I'm studying mechanical engineering at ETH Zurich with the focus on mecatronics an robotics.

Martin (Teammember)

I'm Martin Latta and responsible for controling tasks at the Lunar Robottics Challenge. I'm studying mechanical engineering at ETH Zurich, focussing on mecatronics an robotics in my undergraduate studies.
The main issues on my part will be to generat stable walking gaits (which can also mean crawling or other locomotion modes) an control them during the critical step phases, preventing the robot to tilt over.

Andreas (Teammember)

Hi, I'm Andreas Lauber. In this project, I work on the mechanical design and will take care of the manufacturing organization.
I'm an undergrad student in mechanical engineering with focus on robotics and mechatronics at the ETH Zurich. The main task I'm dealing with is to design an rigid body for the robot to ensure stable walking abilities.



Monday, June 30, 2008

David (Teammembers)

To give a quick introduction of our team, I’d like to introduce myself (hoping the others will follow suit soon).
I'm David Remy. My tasks in the Lunar Robotics Challenge are the coordination of the different work packages and giving support to our undergrad team members. I’ll also be helping with the control issues and I am the official team representative. Since February 2007 I'm doing my PhD at the Autonomous Systems Lab here at ETH Zurich. My research topic is legged locomotion, with a specially focus on the potential of exploiting the periodic nature of legged locomotion to save energy.

Tuesday, June 24, 2008

Ready to roll (or better: walk)

On Tuesday June 10th, we received the great news that our team was invited to the ESA Lunar Robotics Challenge. Starting from now, we have about four and a half months time to develop a robotic device that will decent into a crater on the moon to look for hydrogen rich ore. Scientist believe that in the depths of lunar craters, in the absence of sunlight water might be preserved frozen solid which could be crucial, if one day we would like to establish a permanent base on the moon.

The conditions in such a crater make such an exploration very different from common planetary explorations. The crater edge is extremely steep (about 40 deg) and will be covered by rocks, boulders, and terraces. Landslides can occur every time the sand on the crater's slope is touched. Everything that makes the crater interesting for science (no sunlight, cold temperature, depth) makes it hard to explore.

To cope with this challenge, we proposed a robotic platform with four legs that can walk upright on normal surface, but switches to an arachnoid crawling gait when the terrain becomes rough.

In this Blog, we will keep you updated about the developing process, the steps we're taking, and the setbacks we suffer until we compete with our quadruped against seven other teams in late October.