By Daniel Chan
Directly after our positive and fairly triumphant blog post last Friday, we went back into the long duct of Labarintos to test our robot on the 100 foot cable. Determined to find whatever was at the other end of the duct, we went over quite a few major rocks, and ended up stopping at a point where the robot was too big to fit through, around 80 feet deep into the duct. The way back was less successful, as we got stuck and had to pull the robot with the cable to get it over a few of the big rocks. We started driving it back again when we saw a wheel in the live stream - and realized that a wheel had popped off our robot. After agonizing for a few minutes about if there was anything we could do, we ultimately realized we had to keep going, partially pulling the robot by the cord because of its newfound lack of balance. A few seconds later we saw wheels 2 and 3 in the view, and Jack and I just didn't know what to do.
We ultimately just sadly pulled the robot the rest of the way back. The wheels are stuck about 60 feet into building A - the largest building on the site. Feeling defeated, we hiked up to the top of the hill overlooking the town, and just sat up there for a little while - then we went and ate chocolate cake and hot chocolate at a local café, and began to plan our comeback.
The duct is theoretically big enough that some of the smallest girls on our team could climb in, and our professor has actually been surprisingly supportive of the idea, but neither Jack nor I want to send anyone down the tunnel. It'd take around an hour round trip of an army crawl through a one square foot, dusty and rocky duct, and it's much easier to crawl in than it is to go backwards. So ultimately, we said no to that idea (although thanks to Medora, Alessandra, and Liz for being surprisingly excited and willing to attempt this plan!) The unfortunate part is that these wheels - which worked very well - were not the ones we originally hoped to use, and because of that we a) did not bring backups and b) needed to 3D print adapters in order to use them. In order to be able to get replacements, we not only needed to somehow obtain the wheels, but also our custom adapters. Thanks to the generosity of a lot of our friends back in the Bay Area, we were able to work out a solution. One of John's graduate students is coming to Chavin in about a week, and we were able to ship him a new set of wheels, and get some of Jack's PRL CA friends to 3D print us new adapters and get them to the graduate student! Thanks to Elliot Helms, Thom Trzpit, Dan Somen, Vicky Le, and John Wolf for working through our frantic texts and phone calls, and for Brian Ward and James Wang for offering their support as well.
Friday we started working on new solutions, and planning for what we would do in the meantime as we waited for the new parts. The motors we're using require a press fit, and the only remaining wheels we had that could use that press fit were way too small to be able to get over the large rocks in the tunnels. Our solution ended up being to try to epoxy the smaller wheels into bigger wheels, but we had to take a 2 hour bus ride to the nearest large-ish city in order to get epoxy (or so we thought, this week we found some reasonably strong epoxy at a nearby hardware store). Jack spent the weekend in Huaraz getting epoxy, running errands, and hiking, while Daniel recovered from the Andean flu which took out ~15/18 of our Stanford cohort.
The epoxied wheels solution is actually working surprisingly well, and we've been spending the rest of the week trying to learn and improve the software. We're creating a GUI (graphical user interface) for my computer, and creating a web server on the Raspberry Pi which we can communicate with over the ethernet cable and is more efficient than the previous code. We're also working on integrating the distance sensor, moving our code on the Pi from Java to Python for various reasons, and Jack is working on revamping the Arduino code. We also integrated a new joystick, so we can control the robot with that instead of just keyboard keys! We hope all these improvements will make the robot easier for someone else to use, and overall more reliable. However this has also meant a lot of time just spent in the lab on our computers, and as internet is unreliable, sometimes we run into problems that are very hard to debug.
We tested the fixed-up robot on Wednesday, and were able to go ~70 feet down two more tunnels without issue. No particularly exciting discoveries yet, but we think that we potentially found that one of the ducts connects to another gallery in building A! Most likely a previously discovered duct, as it appears to have some sort of wood scaffolding inside, but we need to make the robot smaller so we can investigate this more thoroughly. Another duct appears to end in a collapse.
In non robot news, we've sung so many songs, played a lot of ukelele and guitar, and mostly recovered from our illnesses! We also watched 4/5 movies of the Twilight Saga, which has led to multiple occasions where people (Liz and Alessandra) have cried from laughter. Liz also played us a number of songs she's written, which are absolutely amazing! The archeology team opened up a brand new gallery with a 1500(ish) year old trash heap which John is just giddy about, and discovered a total of 13 skeletons this week! It's been a really exciting time, and lots of great stuff to distract us from our occasional robot woes. The engineers had the opportunity to help one of the archeologists clean mountains of old Llama and other animal bones, and we were able to go to a science-fair-like day at a local school, where we presented our robot and got kids excited about engineering! They absolutely loved the robot, and it was a great adventure for Jack and I to do a semi-technical presentation in Spanish. Last night was the beginning of the town fiesta, which is a week of partying and fireworks, so I'm sure we'll have many more pictures and stories to share from that next week.
Comments