birth of an Idea.... ok maybe a real refinement....
After my third time looking at the various "dashpc" or car computer setups out there I am finally going to write up a set of papers and a underlying framework for a REAL automotive computing platform that is linux based.
I had back years ago a car stereo called an autopc. it sucked, was slow, would easily lock up requiring you to do fancy things to reboot it and had horribly limited abilities.
but compared to the car computer projects that people are doing today it was completely perfect and worked great!
these abortions that people are working on have a user interface that is worthless for a driver. sorry but tiny on screen buttons that take a person time to figure out where they are is wrong... they are also using expensive touchscreens... I'm looking at using a cheap car tv lcd and a line of 4 buttons on each side of the screen with a IR remote to suppliment input. HUGE fonts, HUGWE graphics and voice synthesis.
this will mean that almost all the apps will have to be custom made. even though there is NO navigation software for linux no matter what the GPS drive people say (it's a moving GPS map NO NAVIGATION IS POSSIBLE!)
so I will be setting out on that road yet once more... mini-itx boards make it more feasable now, and cheap car TV lcd's are dirt cheap now compared to 3 years ago....
Here's my framework..
Linux probably based on a modified slackware.... remove lots of cruft so that busybox is used.... goal is to make it as small and fast as possible.
The GUI will be based on PicoGUI.... yes the lead developer has abandoned it, but it is still the best thing to use for this project.
Flite for the speech system, madlib+madplay for mp3 audio playback, mplayer for video gpsd for the gps data stream and use swsup to suspend to disk when needed.
basically approach the WHOLE thing as a embedded point of view. everyone else is using bloated desktop setups that suck horribly. a vehicle computing platform is SPECALIZED! you dont fricking need an office suite or a complete OS with lots of cruft... let's use as little of that processor as possible people!
I had back years ago a car stereo called an autopc. it sucked, was slow, would easily lock up requiring you to do fancy things to reboot it and had horribly limited abilities.
but compared to the car computer projects that people are doing today it was completely perfect and worked great!
these abortions that people are working on have a user interface that is worthless for a driver. sorry but tiny on screen buttons that take a person time to figure out where they are is wrong... they are also using expensive touchscreens... I'm looking at using a cheap car tv lcd and a line of 4 buttons on each side of the screen with a IR remote to suppliment input. HUGE fonts, HUGWE graphics and voice synthesis.
this will mean that almost all the apps will have to be custom made. even though there is NO navigation software for linux no matter what the GPS drive people say (it's a moving GPS map NO NAVIGATION IS POSSIBLE!)
so I will be setting out on that road yet once more... mini-itx boards make it more feasable now, and cheap car TV lcd's are dirt cheap now compared to 3 years ago....
Here's my framework..
Linux probably based on a modified slackware.... remove lots of cruft so that busybox is used.... goal is to make it as small and fast as possible.
The GUI will be based on PicoGUI.... yes the lead developer has abandoned it, but it is still the best thing to use for this project.
Flite for the speech system, madlib+madplay for mp3 audio playback, mplayer for video gpsd for the gps data stream and use swsup to suspend to disk when needed.
basically approach the WHOLE thing as a embedded point of view. everyone else is using bloated desktop setups that suck horribly. a vehicle computing platform is SPECALIZED! you dont fricking need an office suite or a complete OS with lots of cruft... let's use as little of that processor as possible people!
Comments
Post a Comment