search
This is Spencer Gardner
Email: snppla@gmail.com
Phone: 801-717-0249 (cell/text)
Fall plans
- Work with creating a minimal configuration for the UAV system.
- Specifically work on condensing the video mosiac and the uav navigation software together
- This is coming along. The development environment for the wonder client is very sensitive, but it compiles now. I've spent time understanding the wonder client. Since it and phairwell are written using the QT toolkit, integrating them should be a lot easier.
- MAG: Rather than writing one piece of code that integrates the two pieces (wonder client and phairwell), I think it might be better to just open communication between the two programs through a socket. The two programs are independent and I'm afraid that integrating them will be a headache. Since the amount of info that needs to be communicated between them is small, I'm hopeful that a socket approach will be easier. Of course, we'd have to figure out how to move back and forth between the two interfaces, allowing one interface to be minimized but still display the key information.
- Continue to work on modeling human movement.
- My biggest problem is distinguishing on trail and off trail data. I've been thinking if we use most of the tracklogs from everytrail then perhaps we could find some interesting patterns of how people move.(I think roads and trails have formed because of the way people moved from point a to b. Plus people's decision in which path they take. I think data for how people move off trail in an unfamiliar environment would require field studies)
- MAG: I really think we need the visualization stuff working to solve this. If the vis stuff is working, then we should be able to superimpose trail data with tracklogs and visually inspect what is on-trail and off-trail. I think I know of someone who can do this labor intensive work for us, once we have the vis stuff working.
- Work on the track visualization.
- Instead of trying to write my own interface, I've decided to generate kml files for google earth.
- MAG: Good call!
Random thoughts
- http://www.everytrail.com seems promising. I would like to just write a script to download many of the logs. The trips are arranged into categories (IE geocaching, motorcycling, biking, hiking, etc.) Could provide information of how people move when doing specific activities. Makes me wonder if there's a real need to collect our own logs…
- I've been thinking of ways to analyze data. (Could setup a variety of different scenarios, look for when people are faced with those scenarios (USGS provides a supposedly easy to use elevation query api…) and create statics for those scenarios. (Every possible scenario on even a heptagon would seem too much. Maybe come up with about a dozen different terrain scenarios….)
- Querying elevation data from USGS seems to be painless.
- Querying vegetation density is a possibility. However the data is sparse (and changes depending on the time of year)
- Long term: Think about how to analyze the tracklogs.
- I'm finding that it's really difficult to determine if a tracklog is off trail (Would need a good resource of every trail for a given region. Using google earth and intuitive heuristics seems too inaccurate and time consuming.)
- Would using the mass amount of movement from the tracklogs from everytrail be beneficial in seeing how people move? Even if people are moving on streets or trails, wouldn't terrian still have an impact?
- Walking straight into circles by jan souman provides some possible heuristics for our movement function
