Running on Ice

A lot has happened here at McMurdo since my last post. Our deployment date to the shear zone keeps slipping back, so we're doing more work to test and prepare Yeti on the ice around the station. Two days ago, Jim and I took Yeti out to the SPoTSA (South Pole Traverse Staging Area), and set up our first endurance course to run Yeti for several miles. It had already performed just find on several shorter courses closer to home, so we were both expecting this to be a bit of a formality. We set everything up and pressed play, and Yeti drove off into the sunset towards the outbound waypoint. After about 1km, it hit the next two waypoints and turned to head home, but then kept turning in a 50m circle, and kept going around and around. This was pretty bizarre, but radio comms had been dropping out, so we set it up again and chased it with a truck to stay in radio range. The exact same thing happened again, and left us head scratching.


Yeti outbound on the endurance tests with Mt. Discovery in the background.

We headed back in and I downloaded the telemetry data. It became clear that a bug in the error handling for the GPS input had caused the problem, and the GPS had dropped out at exactly the same point in both runs. There appears to be an intermittent power connection to the GPS which causes some brief dropouts, and the program caused the robot to hold its last command, which in this case was to turn slightly to the right, making it drive in large circles indefinitely.

I rewired the power distribution and tightened down all of the connections, and added a much more robust error handling routine to the code in a late-night push to get everything fixed and set up for another day of testing. I managed to have a bit of a nailbiter of a night, however, when I was tested the intermittent power failures and managed to flip Yeti over, snapping the GPS antenna off. Luckily, everything other than the pole holding the antenna up was fine, and the machine shop at vehicle maintenance was able to build a MUCH nicer antenna mount in a few hours. I then spent yesterday re-running endurance tests, and everything worked out extremely well, setting us up for our first runs with the radar units.


I saw a 'Sun Dog,' while performing endurance tests. These are like rainbows but formed due to small snow crystals in the air, and are all white.

There are several interesting applications apart from crevasse detection that we're going to test Yeti out on. We're going to try looking at sea ice to determine thickness and quality to see if it's feasible to use the robot to survey the airfields here. Apparently they have to drill 30 boreholes a day to profile the ice conditions, and we could dramatically improve the efficiency of that process with Yeti if the radar has a high enough resolution. Allan Delaney will be helping us with the radar to test this and see if it works, which could potentially open up another source for grant funding if successful!


A C-17 on the ice runway, viewed from SPoTSA