[s-cars] Fuzz logic vs adaptive learning

[Keith Maddock]

Linus wrote:
>FWIW, this story was told to impress on me the practical value of math vs. the
>brute-force approach of much of today's programming code.  I wonder what
>lessons our car's s/w engineers may have learned or ignored from these earlier
>times.

(writing from Sweden near the Arctic Circle as my laptop is compiling 
embedded firmware for the ABS/TC/DSC controller of a MY2006 vehicle 
project....)

Writing / Developing / Calibrating embedded passenger vehicle firmware is 
an interesting field to be certain, and I imagine subject to some very 
different constraints than anything related to aerospace and defense.

These days, it ends up being a trade-off of many items.
-  code execution time vs. ROM vs. RAM size
-  code development complexity vs vehicle development/calibration complexity
-  cost of processor speed /ROM/RAM per production unit vs. development 
cost/time

The following items increase the tendency of look-up tables:
  - it seems that processor speed is currently more expensive than ROM and RAM.
  - developing complex vehicle models takes a lot of software development 
time and can be difficult to test 100%
  - developing complex vehicle models with complex coefficients requires a 
lot of vehicle development time and skill
  -  development cycles for vehicles  are getting shorter and shorter
  - vehicle variations are increasing, increasing development cost per 
production unit.
  - vehicle manufactures try to decrease development cost  and time
  - increasing complexity of functionality and integration with other 
vehicle controllers
    (ie ABS controller telling active steering to counter steer when 
braking on uneven surfaces or telling active suspension to change damping)



Cheers,
Keith

keith at maddock.com    http://keith.maddock.com/

93 S4 (mit RS2 + BR upgrades) Portland OR
95 968CS (Ring Tool + Daily Driver) Koblenz, DE
1/8 89 GTI  16V (Spare Ring Toy)  Nürburg, DE