Загрузка страницы

Autonomy Talks - Wilko Schwarting: Learning & Control for Interactions in Mixed Environments

Autonomy Talks - 16/08/2021

Speaker: Dr. Wilko Schwarting, MIT/ISEE AI

Title: Learning and Control for Interactions in Mixed Human-Robot Environments

Abstract: Autonomous robots are on the verge of transforming our homes, factories, and roads. But to reap the tremendous benefits that robots offer to society, we must ensure that they can interact with humans seamlessly and safely. Towards this goal , my work focuses on giving robots the ability to autonomously acquire complexbehaviors, apply learned skills interactively considering the predicted intent of other agents, while explicitly modeling uncertainty throughout all levels of planning to ensure graceful recovery from prediction mismatch. In this talk, I will first present intelligent agents that learn how to reason about human behavior and peoples intentions. By incorporating social behavior, we can measure peoples willingness to cooperate when our interests are not aligned and negoti- ate through game-theoretic interactions. Second, I will show how in stochastic environ- ments with partial observability autonomous agents can leverage information gain and reason about others’ beliefs by combining game-theoretic and belief-space planning. I will present fast , scalable, and calibrated uncertainty estimation of neural networks and discuss how we can leverage uncertainty estimates in learned models for safe operation and efficient exploration in RL. Third, to learn complex behaviors, I will present reinforcement learning agents that learn competitive visual control policies through self-play in imagination. They learn complex skills from competition by imagining multi-agent inter- action sequences in the compact latent space ofa learnedworldmodel . These approaches enable autonomous agents to acquire complex skills from interaction, to perform robustly in uncertain environments, and to communicate intent through naturalistic behavior.

If you want to know more, please visit our official webpage: https://idsc.ethz.ch/research-frazzoli/autonomy-talks.html

Видео Autonomy Talks - Wilko Schwarting: Learning & Control for Interactions in Mixed Environments канала Autonomy Talks
Показать
Комментарии отсутствуют
Введите заголовок:

Введите адрес ссылки:

Введите адрес видео с YouTube:

Зарегистрируйтесь или войдите с
Информация о видео
16 августа 2021 г. 20:57:32
01:01:29
Другие видео канала
Autonomy Talks - David Fridovich-Keil: Learning Mixed Strategies in Lifted Trajectory GamesAutonomy Talks - David Fridovich-Keil: Learning Mixed Strategies in Lifted Trajectory GamesAutonomy Talks - Kaidi Yang: Operation of AV-Enabled Future Mobility SystemsAutonomy Talks - Kaidi Yang: Operation of AV-Enabled Future Mobility SystemsAutonomy Talks - Kevin D. Smith: Semicontracting Dynamical SystemsAutonomy Talks - Kevin D. Smith: Semicontracting Dynamical SystemsAutonomy Talks - Neda Masoud: Cybersecurity in Connected and Automated Vehicle SystemsAutonomy Talks - Neda Masoud: Cybersecurity in Connected and Automated Vehicle SystemsAutonom Talks - Arash Ajoudani: Learning and control for interactive robotsAutonom Talks - Arash Ajoudani: Learning and control for interactive robotsAutonomy Talks - Takashi Tanaka: Minimum-Information Kalman-Bucy FilteringAutonomy Talks - Takashi Tanaka: Minimum-Information Kalman-Bucy FilteringAutonomy Talks - Erdem Bıyık: Learning how to Route Autonomous Vehicles on Shared RoadsAutonomy Talks - Erdem Bıyık: Learning how to Route Autonomous Vehicles on Shared RoadsAutonomy Talks - Stephanie Gil: Resilient Coordination in Networked Multi-Robot TeamsAutonomy Talks - Stephanie Gil: Resilient Coordination in Networked Multi-Robot TeamsAutonomy Talks - Nicolò Pagan: Game Theoretical Inference of Human Behavior in Social NetworksAutonomy Talks - Nicolò Pagan: Game Theoretical Inference of Human Behavior in Social NetworksAutonomy Talks - Sanjiban Choudhury: The Virtues of Laziness in Model-based RLAutonomy Talks - Sanjiban Choudhury: The Virtues of Laziness in Model-based RLAutonomy Talks - Nicolas Lanzetti: Dynamic Programming in Probability Spaces via Optimal TransportAutonomy Talks - Nicolas Lanzetti: Dynamic Programming in Probability Spaces via Optimal TransportAutonomy Talks - Abolfazl Lavaei: Automated Verification and Control of Large-​Scale Stochastic CPSsAutonomy Talks - Abolfazl Lavaei: Automated Verification and Control of Large-​Scale Stochastic CPSsAutonomy Talks - Spring Berman: Scalable Control of Robotic Swarms in Uncertain EnvironmentsAutonomy Talks - Spring Berman: Scalable Control of Robotic Swarms in Uncertain EnvironmentsAutonomy Talks - Laura Ferranti: An Optimization-based Framework for Local Robot Motion PlanningAutonomy Talks - Laura Ferranti: An Optimization-based Framework for Local Robot Motion PlanningAutonomy Talks - Yingzhao Lian: New Perspectives on Advanced Building ControlAutonomy Talks - Yingzhao Lian: New Perspectives on Advanced Building ControlAutonomy Talks - Jonathan Gammell: Measuring and Planning Motion in RoboticsAutonomy Talks - Jonathan Gammell: Measuring and Planning Motion in RoboticsAutonomy Talks - Stelian Coros: Physics-based Modeling and the Quest for Intelligent RobotsAutonomy Talks - Stelian Coros: Physics-based Modeling and the Quest for Intelligent RobotsAutonomy Talks - Stephen James: Sample-Efficient Robot LearningAutonomy Talks - Stephen James: Sample-Efficient Robot LearningAutonomy Talks - Tichakorn Wongpiromsarn: Formal Methods for Control Synthesis of Autonomous SystemsAutonomy Talks - Tichakorn Wongpiromsarn: Formal Methods for Control Synthesis of Autonomous SystemsAutonomy Talks - Margarita Grinvald: Dense Object-​Level MappingAutonomy Talks - Margarita Grinvald: Dense Object-​Level Mapping
Яндекс.Метрика