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:20141026T030000 TZOFFSETFROM:+0200 TZOFFSETTO:+0100 RDATE:20151025T030000 TZNAME:CET END:STANDARD BEGIN:DAYLIGHT DTSTART:20150329T020000 TZOFFSETFROM:+0100 TZOFFSETTO:+0200 TZNAME:CEST END:DAYLIGHT END:VTIMEZONE BEGIN:VEVENT UID:calendar.7033.field_data.0@www.diag.uniroma1.it DTSTAMP:20260414T114408Z CREATED:20150416T085222Z DESCRIPTION:Robots that are capable of successfully operating in open envir onments\, often interacting with other robots and humans\, must find ways to cope with uncertainty and incompleteness of information across a hierar chy of scales ranging from the lower levels of sensorimotor signals to mor e temporally extended aspects of autonomous behaviour. I will present resu lts from recent work in my group that address two aspects of this broad ch allenge - defining and learning action-oriented symbols\; and using catego risation over a space of policies to achieve flexible interaction with oth er agents.I will first set the scene by mentioning a few examples of gener ative models that can be used for intention\, activity and choice predicti on\, as applied to robotic systems. I will argue that success of these mod els depends on having good hierarchical representations.Motivated thus\, I will describe a novel algorithm for multiscale classification of trajecto ries in robotic systems. Unlike previous sampling-based approaches in robo tics\, which use graphs to capture information about the path-connectednes s of a configuration space\, we construct a multiscale approximation of ne ighborhoods of the collision free configurations based on filtrations of s implicial complexes. Our approach thereby extracts additional homological information\, which is essential for a topological trajectory classificati on of sets of trajectories starting and ending in two fixed points. Using a cone construction\, we then further generalize this approach to classify sets of trajectories even when start and end points are allowed to vary i n a path-connected subset. We furthermore show how an augmented filtration of simplicial complexes based on an arbitrary function on the configurati on space\, such as a costmap\, can be defined to incorporate additional co nstraints. We evaluate this in up to 6-dim configuration spaces\, in simul ation as well as real world experiments with the Baxter and PR2 robots.Nex t\, I will outline a model for ad hoc multi-agent interaction without prio r coordination\, which utilises categorisation in an explicitly strategic setting. By conceptualizing the interaction as a stochastic Bayesian game\ , the choice problem is formulated in terms of types in an incomplete info rmation game\, allowing for a learning algorithm that combines the benefit s of Harsanyi’s notion of types and Bellman’s notion of optimality in sequ ential decisions. I will quote some results pertaining to the optimality o f this algorithm even when it is based on a type space that is incorrect i n a sense. I will also present preliminary results from experiments involv ing human-machine interaction where we show a better rate of coordination than alternate multi-agent learning algorithms.Time permitting\, I will co nclude with a brief description of some problem domains we are pursuing in our robotics laboratory\, that demand such a hierarchical treatment of un certainty.Robots that are capable of successfully operating in open enviro nments\, often interacting with other robots and humans\, must find ways t o cope with uncertainty and incompleteness of information across a hierarc hy of scales ranging from the lower levels of sensorimotor signals to more temporally extended aspects of autonomous behaviour. I will present resul ts from recent work in my group that address two aspects of this broad cha llenge - defining and learning action-oriented symbols\; and using categor isation over a space of policies to achieve flexible interaction with othe r agents. Dr. Subramanian Ramamoorthy is a Reader (Associate Professor) in the School of Informatics\, University of Edinburgh\, where he has been s ince 2007. He is Coordinator of the EPSRC Robotarium Research Facility\, a nd Executive Committee Member for the Centre for Doctoral Training in Robo tics and Autonomous Systems. Previously\, he received a PhD in Electrical and Computer Engineering from The University of Texas at Austin. He is a M ember of the Young Academy of Scotland at the Royal Society of Edinburgh\, where he co-chairs the Industry Working Group.His current research is foc ussed on problems of autonomous learning and decision-making under uncerta inty\, by long-lived agents and agent teams interacting within dynamic env ironments. This work is motivated by applications in autonomous robotics\, human-robot interaction\, intelligent interfaces and other autonomous age nts in mixed human-machine environments. These problems are solved using a combination of methods involving layered representations based on geometr ic/topological abstractions\, game theoretic and behavioural models of int er-dependent decision making\, and machine learning with emphasis on issue s of transfer\, online and reinforcement learning.His work has been recogn ised by nominations for Best Paper Awards at major international conferenc es - ICRA 2008\, IROS 2010\, ICDL 2012 and EACL 2014. He serves in editori al and programme committee roles for conferences and journals in the areas of AI and Robotics. He leads Team Edinferno\, the first UK entry in the S tandard Platform League at the RoboCup International Competition. This wor k has received media coverage\, including by BBC News and The Telegraph\, and has resulted in many public engagement activities\, such as at the Roy al Society Summer Science Exhibition\, Edinburgh International Science fes tival and Edinburgh Festival Fringe.Before joining the School of Informati cs\, he was a Staff Engineer with National Instruments Corp.\, where he co ntributed to five products in the areas of motion control\, computer visio n and dynamic simulation. This work resulted in seven US patents and numer ous industry awards for product innovation. DTSTART;TZID=Europe/Paris:20150421T140000 DTEND;TZID=Europe/Paris:20150421T140000 LAST-MODIFIED:20150420T131830Z LOCATION:Aula A2 SUMMARY:Abstractions for Robots Interacting with Open Worlds - Dr. Subraman ian Ramamoorthy URL;TYPE=URI:https://www.diag.uniroma1.it/node/7033 END:VEVENT END:VCALENDAR