Learning to Fly like a Bird. Tedrake, R., Jackowski, Z., Cory, R., Roberts, J. W., & Hoburg, W. Processing, 2009. abstract bibtex Birds routinely execute aerial maneuvers that are far beyond the capabilities of our best aircraft control systems. The complexity and variability of the aerodynamics during these maneuvers are formidable, with dominant flow structures (e.g., vortices) that are difficult to predict robustly from first-principles (Navier-Stokes) models. Here we argue that machine learning will play an important role in the control design process for agile flight by building data-driven ap- proximate models of the aerodynamics and by synthesizing high-performance nonlinear feedback policies based on these approximatemodels and trial-and-error experience. This ar- ticle highlights some of the more remarkable characteristics of natures flyers, and describes the challenges involved in replicating this performance in our machines. We conclude by describing our two-meter wingspan autonomous robotic bird and some initial results usingmachine learning to design control systems for bird-scale, supermaneuverable flight.
@article{Tedrake2009,
title = {Learning to {Fly} like a {Bird}},
abstract = {Birds routinely execute aerial maneuvers that are far beyond the capabilities of our best aircraft control systems. The complexity and variability of the aerodynamics during these maneuvers are formidable, with dominant flow structures (e.g., vortices) that are difficult to predict robustly from first-principles (Navier-Stokes) models. Here we argue that machine learning will play an important role in the control design process for agile flight by building data-driven ap- proximate models of the aerodynamics and by synthesizing high-performance nonlinear feedback policies based on these approximatemodels and trial-and-error experience. This ar- ticle highlights some of the more remarkable characteristics of natures flyers, and describes the challenges involved in replicating this performance in our machines. We conclude by describing our two-meter wingspan autonomous robotic bird and some initial results usingmachine learning to design control systems for bird-scale, supermaneuverable flight.},
journal = {Processing},
author = {Tedrake, Russ and Jackowski, Zack and Cory, Rick and Roberts, John William and Hoburg, Warren},
year = {2009},
keywords = {Control theory, Flapping-Wings, Fluid Dynamics, Flying robots, Machine learning, Reinforcement learning},
pages = {1--7},
}
Downloads: 0
{"_id":"fW3JwibHtGgjG3yJ6","bibbaseid":"tedrake-jackowski-cory-roberts-hoburg-learningtoflylikeabird-2009","author_short":["Tedrake, R.","Jackowski, Z.","Cory, R.","Roberts, J. W.","Hoburg, W."],"bibdata":{"bibtype":"article","type":"article","title":"Learning to Fly like a Bird","abstract":"Birds routinely execute aerial maneuvers that are far beyond the capabilities of our best aircraft control systems. The complexity and variability of the aerodynamics during these maneuvers are formidable, with dominant flow structures (e.g., vortices) that are difficult to predict robustly from first-principles (Navier-Stokes) models. Here we argue that machine learning will play an important role in the control design process for agile flight by building data-driven ap- proximate models of the aerodynamics and by synthesizing high-performance nonlinear feedback policies based on these approximatemodels and trial-and-error experience. This ar- ticle highlights some of the more remarkable characteristics of natures flyers, and describes the challenges involved in replicating this performance in our machines. We conclude by describing our two-meter wingspan autonomous robotic bird and some initial results usingmachine learning to design control systems for bird-scale, supermaneuverable flight.","journal":"Processing","author":[{"propositions":[],"lastnames":["Tedrake"],"firstnames":["Russ"],"suffixes":[]},{"propositions":[],"lastnames":["Jackowski"],"firstnames":["Zack"],"suffixes":[]},{"propositions":[],"lastnames":["Cory"],"firstnames":["Rick"],"suffixes":[]},{"propositions":[],"lastnames":["Roberts"],"firstnames":["John","William"],"suffixes":[]},{"propositions":[],"lastnames":["Hoburg"],"firstnames":["Warren"],"suffixes":[]}],"year":"2009","keywords":"Control theory, Flapping-Wings, Fluid Dynamics, Flying robots, Machine learning, Reinforcement learning","pages":"1–7","bibtex":"@article{Tedrake2009,\n\ttitle = {Learning to {Fly} like a {Bird}},\n\tabstract = {Birds routinely execute aerial maneuvers that are far beyond the capabilities of our best aircraft control systems. The complexity and variability of the aerodynamics during these maneuvers are formidable, with dominant flow structures (e.g., vortices) that are difficult to predict robustly from first-principles (Navier-Stokes) models. Here we argue that machine learning will play an important role in the control design process for agile flight by building data-driven ap- proximate models of the aerodynamics and by synthesizing high-performance nonlinear feedback policies based on these approximatemodels and trial-and-error experience. This ar- ticle highlights some of the more remarkable characteristics of natures flyers, and describes the challenges involved in replicating this performance in our machines. We conclude by describing our two-meter wingspan autonomous robotic bird and some initial results usingmachine learning to design control systems for bird-scale, supermaneuverable flight.},\n\tjournal = {Processing},\n\tauthor = {Tedrake, Russ and Jackowski, Zack and Cory, Rick and Roberts, John William and Hoburg, Warren},\n\tyear = {2009},\n\tkeywords = {Control theory, Flapping-Wings, Fluid Dynamics, Flying robots, Machine learning, Reinforcement learning},\n\tpages = {1--7},\n}\n\n","author_short":["Tedrake, R.","Jackowski, Z.","Cory, R.","Roberts, J. W.","Hoburg, W."],"key":"Tedrake2009","id":"Tedrake2009","bibbaseid":"tedrake-jackowski-cory-roberts-hoburg-learningtoflylikeabird-2009","role":"author","urls":{},"keyword":["Control theory","Flapping-Wings","Fluid Dynamics","Flying robots","Machine learning","Reinforcement learning"],"metadata":{"authorlinks":{}},"html":""},"bibtype":"article","biburl":"https://bibbase.org/zotero/girdhar","dataSources":["XFrKPG99s5t3W7xuW"],"keywords":["control theory","flapping-wings","fluid dynamics","flying robots","machine learning","reinforcement learning"],"search_terms":["learning","fly","bird","tedrake","jackowski","cory","roberts","hoburg"],"title":"Learning to Fly like a Bird","year":2009}