ECUs

[Mark Boxsell]

>Hello,
>  Here's a newbie thinking about the obvious.
>
>  In control systems you there are three main categories: stochastic,
>derived and hybrids.  Stochastic systems analyse input and out and and
>try and set up discrete mappings according to optmisation criteria.  
>Derived attempt to solve the physical systems and so specify them as 
>equations rather than tables.  A hydbrid uses both where appropriate.
>Using empirical data to calibrate derived equations and pure
>stochastic solutions where the equations are too difficult.
>
>  It seems, from the discussion here, most ECUs seem to be
>stochastic.  I assume this is due to the "try it and see" empirical
>background of automotive engineers.
>
>  My soldering skills are best not talked about in polite circles so
>an ideal development platform would be one like.
>
>    - ECU running a multitasking operating system
>    - Services available to access input and manipulate outputs (like
>    timing, etc).
>    - Each service would have a local butter to store/read data
>    cyclicly so the optimization can be done externally.
>    - Persistent code/data storage such as FRAM chips.
>    - Laptop interface
>
>
>  So you'd have some, say Java, firmware on the FRAM and more bulky
>code on the PC.  Adding new drivers would create new Java objects to
>manipulate them.
>
>    Does anything close to this exist or is it all pipe dreams?
>
>-- 
>  _--_|\             -|-    Gavin Walker
> /      \             |     
> \_.--.*/ <- Canberra,      gavin.walker at dit.csiro.au
>       v     Australia      Phone +61-6-2167030     Fax +61-6-2167111
>

Gavin,Gavin,Gavin,
                  remember an engine is just a dirty smelly thing which does
not liked being "modelled". 

                 regards,
                         Mark Boxsell.
                         MRB Design