RI Seminar: Russ Tedrake: Robust motion planning for walking robots and robotic birds
Robust motion planning for walking robots and robotic birds
Russ Tedrake
Associate Professor, MIT
October 28, 2011
Abstract
Building robots that can walk or fly with the dexterity of an animal is hard. Our best models of these robots are complicated because they are nonlinear and underactuated, and because the robots are fundamentally subjected to large sources of uncertainty relative to their control authority - for instance, large disturbances like a gust of wind and/or model uncertainty because they are locomoting on unknown terrain. While the field of robotics has seen some success from "intuitive" controllers based on simple models, I believe that a more algorithmic approach is required to scale much beyond the complexity and performance of today's robots.
In this talk, I will demonstrate that combinations of randomized motion planning and trajectory optimization can produce open-loop trajectories for models of the required complexity (nonlinear, hybrid, underactuated). But when one goes to implement these on the real systems, they inevitably fail, in part because trajectory motion planning by itself doesn't provide any language for talking about robustness. Therefore, I'll describe tools from robustness analysis (both linear and nonlinear), feedback planning, and planning in belief space that can help close this gap, and produce solutions that work on real systems. These ingredients can be added on top of most existing motion planners, and can make a substantial difference when working with high-performance dynamic robots.
Speaker Biography
Russ is the X Consortium Associate Professor of Electrical Engineering and Computer Science and Aeronautics and Astronautics at MIT, and a member of the Computer Science and Artificial Intelligence Lab. He is a recipient of the NSF CAREER Award, the MIT Jerome Saltzer Award for undergraduate teaching, the DARPA Young Faculty Award, and was named a Microsoft Research New Faculty Fellow.
Russ received his B.S.E. in Computer Engineering from the University of Michigan, Ann Arbor, in 1999, and his Ph.D. in Electrical Engineering and Computer Science from MIT in 2004, working with Sebastian Seung. After graduation, he joined the MIT Brain and Cognitive Sciences Department as a Postdoctoral Associate. During his education, he has also spent time at Microsoft, Microsoft Research, and the Santa Fe Institute.
Видео RI Seminar: Russ Tedrake: Robust motion planning for walking robots and robotic birds канала cmurobotics
Russ Tedrake
Associate Professor, MIT
October 28, 2011
Abstract
Building robots that can walk or fly with the dexterity of an animal is hard. Our best models of these robots are complicated because they are nonlinear and underactuated, and because the robots are fundamentally subjected to large sources of uncertainty relative to their control authority - for instance, large disturbances like a gust of wind and/or model uncertainty because they are locomoting on unknown terrain. While the field of robotics has seen some success from "intuitive" controllers based on simple models, I believe that a more algorithmic approach is required to scale much beyond the complexity and performance of today's robots.
In this talk, I will demonstrate that combinations of randomized motion planning and trajectory optimization can produce open-loop trajectories for models of the required complexity (nonlinear, hybrid, underactuated). But when one goes to implement these on the real systems, they inevitably fail, in part because trajectory motion planning by itself doesn't provide any language for talking about robustness. Therefore, I'll describe tools from robustness analysis (both linear and nonlinear), feedback planning, and planning in belief space that can help close this gap, and produce solutions that work on real systems. These ingredients can be added on top of most existing motion planners, and can make a substantial difference when working with high-performance dynamic robots.
Speaker Biography
Russ is the X Consortium Associate Professor of Electrical Engineering and Computer Science and Aeronautics and Astronautics at MIT, and a member of the Computer Science and Artificial Intelligence Lab. He is a recipient of the NSF CAREER Award, the MIT Jerome Saltzer Award for undergraduate teaching, the DARPA Young Faculty Award, and was named a Microsoft Research New Faculty Fellow.
Russ received his B.S.E. in Computer Engineering from the University of Michigan, Ann Arbor, in 1999, and his Ph.D. in Electrical Engineering and Computer Science from MIT in 2004, working with Sebastian Seung. After graduation, he joined the MIT Brain and Cognitive Sciences Department as a Postdoctoral Associate. During his education, he has also spent time at Microsoft, Microsoft Research, and the Santa Fe Institute.
Видео RI Seminar: Russ Tedrake: Robust motion planning for walking robots and robotic birds канала cmurobotics
Показать
Комментарии отсутствуют
Информация о видео
Другие видео канала
PhD in USA/Canada | Important Application Process Details - All you need to knowRobot building and barefoot running with MIT's Russ Tedrake"New Connections between Motion Planning and Machine Learning" - Sidd SrinivasaFeedback Motion Planning via Sums-of-Squares Verification (Russ Tedrake)RI Seminar: Avik De : Design and control of insect-scale bees and dog-scale quadrupedsRobotic challengerRI Seminar: Mac Schwager : Enabling Robots to Cooperate & Compete: Distributed Optimization...Advice for Young People | Russ Tedrake and Lex FridmanRI Seminar: Raquel Urtasun : A future with affordable Self-driving vehiclesRI Seminar: Melisa Orta Martinez: Design and Analysis of Open-Source Educational Haptic DevicesHow to Shrink Big DataPerching with a Simple Fixed-Wing GliderTutorial: Gait and Trajectory Optimization for Legged RobotsIntroducing TRI Manipulation for Human-Assist RobotsRI Seminar: Kevin Lynch : Robotics and BiosystemsRI Seminar: Robert D. Gregg IV : kinematic/energetic design/robots for agile human locomotionMIT Mini Cheetah Actuator TeardownTeam MIT at the DARPA Robotics ChallengeRI Seminar: Herman Herman : Robots “R” Us: 25 years of NREC