Is there a domain that combines Embedded system design and Control system theory?

[tutankhamen]

During my third year I studied control systems and was quite fascinated with the concept of feedback and controlled output for an input. I modelled few systems and started exploring applications.
I am currently in my fourth year and learning to program electronics aka embedded software development to get hired. For my final year project I am thinking about developing a robot where we do not have to focus much on the mechanical design part and combine both embedded and control theory. Please provide with your insights and also mention some existing projects to learn from.

 

[KeepItSimpleStupid]

Search for PID and fuzzy logic for starters. I ended up with a nice book. Although the concepts are easy, I took PID beyound the concepts.
I didn't implement auto-tuning.

In one control case, I had an environmental chamber with an integral chamber sensor and a surface RTD sensor My -100 to 100% was the proportial band of the OEM controller or +-10 degrees of setpoint. so I could now control the surface temperature by manipulating the chamber setpoint.

 

[crutschow]

Here's a good reference for Fuzzy Logic from Seattle Robotics.

 

[KeepItSimpleStupid]

The first PID system I did was in the 1980's on a mini-computer running FORTRAN and the RT11 OS. We has 7 PID loops, with sequencing and logging. We could control voltage, current, power, and temperature. One was designated the controlled variable and the others acted as limits. I created "stablity creiteria" which would include, open, heating, cooling, stable, conditionally, stable, heat up energy limit exceeded. The later would detect a shorted or misplaced thermocouple on heat-up. It was really, really cool.

It had the ability to create spreadsheet output (never implemented), but spreadsheets were not invented yet. The hook was there.

Basically we ran out of memory even with overlays. Software and true hardware thermocouples were not designed from the beginning because of some cost issues. The thermocouple modules had to be configured in groups of 4 and could be J,K, R or S. Ideally, the system should have handled a type "C" thermocouple. It was used for a long time for one part of the research process.

We wanted to run RSX-11 but management said no because we had no experience.

After we finished the "proof of concept" design, operators wrote programs in FORTRAN that would generate the TXT recipe files.

The was on a VT100 terminal of 80col x 25 lines.

 

[tutankhamen]

I am starting with fuzzy logic. Thanks for your guidance