Tuesday, October 12, 2010

Long Hours and Debugging

The new schedule has proved to be challenging but effective for getting a lot done with Yeti. Yesterday I went to bed at 5am and was supposed to sleep until 2pm, but found that I keep waking up far earlier and end up getting far too little sleep. Despite this, I've had some solid long blocks of time to debug issues with Yeti.

I've been working through a lot of reliability and performance issues that we've seen all along. When we first built the robot, the code and electronics were all developed somewhat ad-hoc, somewhat against the advice of Professor Ray. We've found numerous small reliability issues that are each small, but when compounded, mean that whoever is operating the robot really needs to understand everything that's going on internally. The goal of being here this year is to make a compelling enough case to propose additional funding to develop a more reliable and field-hardened system.

One of the most challenging aspects of developing the code on Yeti is that I actually have to run it around outside in order to test the GPS navigation. When things aren't working, it can be really challenging to figure out exactly why they're not, and the time to iteratively test and debugging code in little pieces is much much higher than when you're writing code to execute on your own computer. I spent the last day building telemetry into the code, so I can get data back over the radio and not have to chase around the robot while holding a laptop that's plugged into it in -20 degree temperatures and high wind. We're really close at this point, and I've made huge improvements in the numerical stability of our navigation code, which I'm hoping to test tomorrow after crevasse rescue training.

During our happy camper survival training, I quickly became annoying in holding down one side of a debate that's always bothered me. If anyone can point out something I'm missing, let me know. The debate was whether if you're cold, you should pee, because your body is expending energy to keep the urine warm. I argued that this is totally wrong, and that it makes sense to pee because it's uncomfortable, but that you don't lose any more heat. Three PhDs in various fields of science, along with a recent graduate who had 'just taken thermo' all argued the other side. The basis of my argument is that all of your heat loss is through your skin via convection, conduction, radiation, and some evaporation, and all of those remain constant before and after urinating (ignoring the negligible change in skin surface area accompanying the smaller volume of your body after peeing). Similarly, since the urine is already at body temperature, you don't need to do any work to change its temperature. If you work out the energy balance, the metabolic work going into maintaining body temperature would equal the m*c*delta T of the urine minus the heat lost on your skin. Since delta T is zero, you're not doing any work to keep it warm. The counter point, as far as I could tell, was simply that "with more mass, your body needs to expend more energy to keep it warm." In reality, the other side didn't really care that much, but it's just so hard to let someone be wrong! Following that line of thought, feel free to correct me: etrautmann@gmail.com

We're heading out to the shear zone on Friday and I likely will be off the grid for a while then. I'll try to post updates when I can.


Post a Comment

Subscribe to Post Comments [Atom]

<< Home