Virtual birds

A few months ago I gave a talk at the ASAB Easter conference in Sheffield about how I’m analysing the foraging rafts using collective behaviour. Of course, what with one thing and another, the work wasn’t as far along as I would have liked. After I got back from fieldwork and crunched the new data, I decided it was time to look at this work again.

One of the main reasons for pushing this work is that I am giving another talk on this at an international conference. In New York. In a few weeks. Do not let my restrained tones hide how terrified/excited I am about this.

The conference is ISBE, the Internation Society of Behavioural Ecologists. I applied for a talk back at the beginning of the year and several months later, here we are:

program

 

(Last talk of the day, hopefully some people will be awake)

I’d really better get some results together hadn’t I?

To recap what I had already done:

-Extracted positions of individual birds from video.

vlcsnap-00028-Tracked these individuals and then used these positions to create trajectories.

– Ran these trajectories through a correction matrix, to account for any distortions the camera might introduce

– Created code to remove unrealistic tracks.

– Wrote further functions that I can use to manually delete and merge tracks.

– Extracted the dive and surface events

dsplot

– Created various graphing functions, such as taking the corrected trajectories and plotting them back to their original positions to check my working.

vlcsnap-00030

What I have done in the last week (working frantically):

– Created a simulation based on zonal interaction models. Virtual birds that mooooooove.

– Wrote code that will compare the real data with my models, using things like radial density.

relpos– Added the ability for my virtual birds to DIVE.

– Modified simulation so the birds can react to these dive and surface events.

What am I doing now?

So the goal now is to produce results by choosing the simulation that best matches up with the data. In order to do this I need to choose the best “weight” for each of the things that might affect a birds movement (the effect of repulsion from other birds vs. the effect of attraction to other birds for example). I am currently exploring these weights manually, but eventually I need to run some code to optimise each of these weights for each piece of data.

I’d also like to add some more complicated diving rules (At the moment diving is totally random) where they will be influenced by other dives. I then need to come up with ways of checking this method of diving against the data, to determine how realistic this sort of behaviour is..

I fly out to New York in a few weeks. This is why I spend my weekends locked in the office.

Just to finish up here is a video from one of my early simulations with some random parameters. Those birds can’t seem to get away from each other fast enough! I suspect they might also need a bit more autonomous movement to stop them doing the crazy little dance when they can’t find other birds.

 

 

Advertisements

5 responses to “Virtual birds

  1. Pingback: Pre-Manhattan Madness | The Shag Project

  2. Pingback: So New York is now a Place I have Been. | The Shag Project

  3. Pingback: More virtual birds | The Shag Project

  4. Pingback: The Shag Project

  5. Pingback: So what do you do after you hand in your thesis anyway? | The Shag Project

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s