Zum Inhalt springen

Benutzer:Dirk Hünniger/plan

Aus Wikibooks
Proposed Experiments on Linux Kernel Development
a overview of suggest changes to the imaging system.
Dirk Hünniger , Gordon Freeman
Wikibooks


Shahzad did do a good job and particularly came up with a working system within 8 month. To build the current setup he needed 2 month. I would like to make a few experiments. If they turn out as expected I would like to start from scratch. Thus throwing away two month of work.

Planned System Setup:

  • Android Phone with jailbreak running Debian.
  • Data recored on USB Stick connected to it.
  • Camera Connected to it via USB. Camera without computing capabilities.

Advantages:

  • Our code is written for Linux. In the current setup it is written for the camera. And if the camera is not available anymore we have to rewrite our code.
  • The size of the camera is much smaller
  • The camera does produce much less heat. Likely at least since it is probably the DSP that is generating most of the heat.
  • We can bend the cable between the camera and the android phone as we like
  • The power consumption will be significantly lower
  • We don't need any IMU anymore. Saves us a lot of money
  • We can by a much cheaper camera.

Expected Problem:

  • Will it work with the phone we got.
  • In order to do real time on Linux we need to write kernel code. Development is very slow since if the code crashed we will have to restart the machine. I might also hose the phone in that. We maybe its not really that dangerous, but you never know.
  • Will an open source IMU running on Android be good enough to work with the post processing pipeline.

Planned Experiments:

  • write a hello world kernel module and make it run.
  • Set up a software IMU in the phone. And try to get an Idea in the precision.
  • Experiment on the realtime issues on the phone