{"674613":{"#nid":"674613","#data":{"type":"event","title":"Ph.D. Dissertation Defense - Niranjan Kannabiran","body":[{"value":"\u003Cp\u003E\u003Cstrong\u003ETitle\u003C\/strong\u003E\u003Cem\u003E:\u0026nbsp; \u003C\/em\u003E\u003Cem\u003ELearning interactive robot behaviors using deep reinforcement learning\u003C\/em\u003E\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u003Cstrong\u003ECommittee:\u003C\/strong\u003E\u003C\/p\u003E\r\n\r\n\u003Cp\u003EDr. Irfan Essa, CoC, Chair, Advisor\u003C\/p\u003E\r\n\r\n\u003Cp\u003EDr. Sehoon Ha,IC, Co-Advisor\u003C\/p\u003E\r\n\r\n\u003Cp\u003EDr. Sonia Chernova, IC\u003C\/p\u003E\r\n\r\n\u003Cp\u003EDr. Justin Romberg, ECE\u003C\/p\u003E\r\n\r\n\u003Cp\u003EDr. Wenhao Yu, Google\u003C\/p\u003E\r\n\r\n\u003Cp\u003EDr. Animesh Garg, IC\u003C\/p\u003E\r\n\r\n\u003Cp\u003EDr. Mark Davenport, ECE\u003C\/p\u003E\r\n","summary":"","format":"limited_html"}],"field_subtitle":"","field_summary":[{"value":"\u003Cp\u003EPerception and interaction form the foundational pillars of every robotics system. The robot learning research community has extensively focused on building control policies informed by the perceptual capabilities of robots. However, two significant challenges hinder the practical application of \u0026nbsp;robots in diverse real-world scenarios.\u0026nbsp;\u003Cbr \/\u003E\r\nFirst, robots frequently lack awareness of the true state of the world, such as the properties of objects they interact with or the structure of a cluttered scene, leading to suboptimal or erroneous interactions.\u003Cbr \/\u003E\r\nSecond, designing control policies for high-DOF robots remains a daunting task. While learning-based approaches offer a promising alternative to model-based control, the extensive reward engineering required to produce policies that reliably transfer to real-world is often impractical due to the complexity and variability of the system.\u0026nbsp;\u003Cbr \/\u003E\r\nIn my PhD research I developed learning-based control policies to improve a robot\u0027s understanding of its environment and solve interactive tasks while circumventing the need for complex reward engineering.\u003C\/p\u003E\r\n","format":"limited_html"}],"field_summary_sentence":[{"value":"Learning interactive robot behaviors using deep reinforcement learning "}],"uid":"28475","created_gmt":"2024-05-09 09:01:20","changed_gmt":"2024-05-09 09:03:19","author":"Daniela Staiculescu","boilerplate_text":"","field_publication":"","field_article_url":"","field_event_time":{"event_time_start":"2024-05-13T14:00:00-04:00","event_time_end":"2024-05-13T16:00:00-04:00","event_time_end_last":"2024-05-13T16:00:00-04:00","gmt_time_start":"2024-05-13 18:00:00","gmt_time_end":"2024-05-13 20:00:00","gmt_time_end_last":"2024-05-13 20:00:00","rrule":null,"timezone":"America\/New_York"},"location":"CODA C1015 Vinings","extras":[],"related_links":[{"url":"https:\/\/gatech.zoom.us\/j\/97282602852?pwd=L0o5VGdYWjI2STZkbU8vMGtqSUpNZz09","title":"Zoom link"}],"groups":[{"id":"434381","name":"ECE Ph.D. Dissertation Defenses"}],"categories":[],"keywords":[{"id":"100811","name":"Phd Defense"},{"id":"1808","name":"graduate students"}],"core_research_areas":[],"news_room_topics":[],"event_categories":[{"id":"1788","name":"Other\/Miscellaneous"}],"invited_audience":[{"id":"78771","name":"Public"}],"affiliations":[],"classification":[],"areas_of_expertise":[],"news_and_recent_appearances":[],"phone":[],"contact":[],"email":[],"slides":[],"orientation":[],"userdata":""}}}