Abstract: In this talk, I will present how I tackled, during my research projects, some of the activities encountered when designing a control system. Designing a control system implies a number of options, decisions and compromises which depend on the properties of the controlled system and on the required performances. To design a control system, the modelling, controller design, analysis, simulation, implementation and verification steps, which represent research directions by themselves, are often required.
I applied an alternative old approach of deriving optimal control laws such as inverse problems of variations calculus or inverse optimal control (IOC) which arouse a renewal of interest among researchers during the years. I addressed inverse optimal control and inverse optimization approaches based parametrized Lagrangians, which reduces the nonlinear parametric optimization problems to linear least square optimization, hence being easier to solve. I used IOC approach to analyse and propose simply but efficient bio-inspired models for redundant biological rhythmical motions, such as: arm motion, postural sway-and-balance, fast-and-normal human locomotion and squat movements. Moreover, I used this IOC approach as a tool to discriminate and reproduce the human fast-and-normal gait for patients with Parkinson’s disease (PD). By combining different research topics such as: estimation, least square methods and interval analysis, I propose a novel IOC method solved in a bounded-error framework was proposed.
Recently, I addressed the navigation planning level for the autonomous navigation of mobile robots and I proposed and validated on a easy to reproduce robotic platform, a new optimal (in terms of path length) and a robust motion planning algorithm based incremental sampling-based motion algorithm, i.e. Rapidly-exploring Random Tree (RRT) and its optimal versions based interval analysis tools.