BEGIN:VCALENDAR VERSION:2.0 PRODID:-//Date iCal//NONSGML kigkonsult.se iCalcreator 2.20.2// METHOD:PUBLISH X-WR-CALNAME;VALUE=TEXT:Eventi DIAG BEGIN:VTIMEZONE TZID:Europe/Paris BEGIN:STANDARD DTSTART:20241027T030000 TZOFFSETFROM:+0200 TZOFFSETTO:+0100 TZNAME:CET END:STANDARD BEGIN:DAYLIGHT DTSTART:20250330T020000 TZOFFSETFROM:+0100 TZOFFSETTO:+0200 TZNAME:CEST END:DAYLIGHT END:VTIMEZONE BEGIN:VEVENT UID:calendar.28643.field_data.0@www.diag.uniroma1.it DTSTAMP:20260403T175810Z CREATED:20250114T102325Z DESCRIPTION:Dr. Paolo Robuffo GiordanoCentre National de la Recherche Scien tifique (CNRS) Rainbow Team\, Irisa and Inria Bretagne Atlantique Rennes\, France Title: Intrinsic Robust Planning for Uncertain RobotsAbstract: Unc ertainties in a robot/environment can have many sources and can be modeled at different levels of detail. For instance\, a typical choice in many ro botics applications is to consider a generic additive “process noise” (or “actuation noise”)\, often taken as Gaussian random variables. This additi ve term can capture the presence of some uncertainty in the robot’s motion /actuation\, but its genericity does not allow\, in general\, to obtain pr ecise predictions of how the actual uncertainties will affect the robot’s motions or actions.However\, in many cases of interest  a model of the rob ot/environment can be considered available with the main source of uncerta inty lumped in the inaccurate knowledge of some model parameters. In these  cases\, the uncertainty is not generic but it has a very specific structu re which\, if exploited\, can lead to better predictions of its effects on the robot motion.In light of these considerations\, in this talk we will review the recent notion of 'closed-loop state sensitivity matrix': this q uantity locally captures how deviations in the model parameters (w.r.t. th eir nominal values) affect the evolution of the robot/environment states i n closed-loop\, i.e.\, by also taking into account the strengths/weaknesse s of the particular control action chosen for executing the task. A norm o f the state sensitivity can\, for instance\, be minimized for generating r eference trajectories that result by construction minimally sensitive to p arametric uncertainties\, thus increasing the intrinsic robustness of thei r tracking in closed-loop.Furthermore\, it is also possible to leverage th e sensitivity matrix to obtain time-varying bounds (tubes) on the state an d/or input evolution assuming a (known) range of variation for the paramet ers. This makes it possible to plan robust trajectories that\, at least lo cally\, ensure the feasibility of the resulting motion (against state/inpu t constraints) also in the presence of parametric uncertainties.The talk w ill revisit the formalization of the closed-loop sensitivity and derived q uantities\, and will illustrate how these notions can be applied to offlin e and online robust trajectory planning for several robotic case studies o f interest (drones\, manipulators).io: Biosketch: Paolo Robuffo Giordano i s a CNRS senior research scientist\, head of the Rainbow group at IRISA/In ria\, Rennes\, France. He holds a PhD degree in Systems Engineering obtain ed in 2008 at the Sapienza University of Rome. From January 2007 to July 2 007 and from November 2007 to October 2008\, he was a research scientist a t the Institute of Robotics and Mechatronics\, German Aerospace Center (DL R)\, Germany\, and from October 2008 to November 2012 he was a senior rese arch scientist at the Max Planck Institute for Biological Cybernetics and scientific leader of the group “Human-Robot Interaction”. His scientific i nterests include motion control for mobile robots and mobile manipulators\ , visual control of robots\, active sensing\, bilateral teleoperation\, sh ared control\, multi-robot estimation and control\, aerial robotics. DTSTART;TZID=Europe/Paris:20250122T120000 DTEND;TZID=Europe/Paris:20250122T120000 LAST-MODIFIED:20250114T120237Z LOCATION:Aula Magna SUMMARY:Seminar of Paolo Robuffo Giordano: Intrinsic Robust Planning for Un certain Robots - Dr. Paolo Robuffo Giordano URL;TYPE=URI:http://www.diag.uniroma1.it/node/28643 END:VEVENT END:VCALENDAR