Notes from my graduate studies at the University of Toronto in the Department of Computer Science.

The crop planning problem for market gardens

Saturday, April 9, 2011

I'm so happy that the Journal of Computers and Electronics in Agriculture exists; it is a wonderful combination of three topics that interest me.  The papers tend to be about sensing and identification of animal or plant features, modelling and planning of agricultural systems, robotics, and decision-support tools.  The papers themselves aren't usually my cup of tea though: most of the papers are aimed at solving problems that are relevant only to large or industrial agricultural operations, or for very specialized applications.  But much of the technology proposed could, theoretically, be adapted and used with the kinds of farming I am interested in: small-scale, mixed farms and market gardens. In any case, I've been following this journal for a while now and I find it to be a fascinating read.  It's especially fun when I run across papers like this, or this.

Anyhow, in order to prepare for my final assignment in Climate Informatics, I spent Friday afternoon scanning through the last five years of the publication ostensibly looking for articles to do with crop planning. (But, I also took the opportunity to look for any other interesting and inspiring articles -- hence the exhaustive search).  The assignment for the class involves writing a research proposal and I am considering writing one to investigate the computational problem of finding an optimal cropping plan and rotation for a market-garden (I'm inspired by some of the work done by Institute for Computational Sustainability on equally unlikely computer science topics) .  The crop planning problem could turn out just to be an annoyingly complex optimization problem without any research value, but I thought I'd check to see what prior work exists before I dismiss it.

Sadly, crop planning isn't something written too much about it in this journal.   Here are the two relevant papers I found (the second isn't actually from CEA, but I found it by following references):
R. Sarker and T. Ray, "An improved evolutionary algorithm for solving multi-objective crop planning models," Computers and Electronics in Agriculture, vol. 68, no. 2, pp. 191-199, Oct. 2009. 
J. Bachinger and P. Zander, "ROTOR, a tool for generating and evaluating crop rotations for organic farming systems," European Journal of Agronomy, vol. 26, no. 2, pp. 130-143, Feb. 2007. 
The Sarker and Ray paper gets close to what I'm looking for actually, and has a wealth of references I could follow up with.  The Bachinger and Zander article is much more focused on organic agriculture as it emphasizes the time dimension in the problem definition (rotating where crops are planted in a field over time is done to avoid nutrient depletion and pest problems -- this is not as big of a problem when your are using fertilizer and spraying pesticides).   At the very least it shows that the topic is research-worthy in some field -- though I'm not yet sure it's worthy of computer science research.

Side note: In doing this review I was reminded of what is probably the most interesting and potentially world-changing (IMHO) applications of computers and electronics in agriculture: robotic weeding.  Think about it.  If we were able to develop effective precision robotic weeding systems the non-organic growers could all but give up herbicides.  That's huge.  And since weed control is such a big job for organic growers, robotic weeding devices means organic growers could do their jobs much more efficiently, and with less tractor work and backache.


Neil said...

Welcome to the world of requirements engineering Jon!

Optimizing multiple criteria in designing a system is something I'm working on.

The agriculture problem comes up a lot in linear optimization (simplex etc). As long as you have all the constraints it should be easy to solve.

Of course, the interesting problems are never that simple.

About Me