Control projects during my academic study ended-up with simplest PID algorithm tuned through successive trials. Code implementation for embedded systems requires an important effort. Time left for modeling, identification and simulation is often null.
I started in 2005 developing a blockset for Matlab/Simulink targeting dsPIC 33F microcontrollers to overcome such constraint. Without a budget for dSPACE, SpeedGoat or another NI alternative, rapid prototyping became possible on any dsPIC based board.
$$ \text{One push button} \left\{ \begin{array}{l} \text{- C code generation,} \\\ \text{- Compilation,} \\\ \text{- Upload & run on the target.} \\\ \end{array} \right. $$
This Model Based Design (MBD) approach enabled efficient research on signal processing validated with real robot for my PhD thesis on bio-robotics from 2006 to 2009. It reduces the time from the simulation to our autonomous robot and replaced somehow our dSPACE platform which was not embedded anyway.
I used the same Rapid Control Prototyping (RCP) to develop data fusion algorithm on motion analysis for wearable sports article at MOVEA in 2010-2011.
I joined Microchip Technology where I am pursuing the development of the MPLAB blockset to target dsPIC and PIC32 microcontrollers.
For company, scientists, and students, rapid prototyping enables focusing on new ideas rather than getting into the details of embedded programming. Shortening the loopback $\lbrace Simulation \Leftrightarrow Hardware \rbrace$ allows improving algorithms, obtaining better results, and reducing the time to market.
Most projects use the free Microchip blockset targeting dsPIC, PIC32 and few others micro-controllers. Old project are available on my old website.
Ph.D Automatic, Signal Processing & Aerial Robotics, 2009
National Center for Scientific Research (CNRS) / University Montpellier II
Master of Research - Signal Processing and Digital Communication, 2004
University of Nice Sophia-Antipolis
, 2004
ESIEE Paris