2026-09-19T13:28:51Zhttps://dspace.mit.edu/server/oai/request

oai:dspace.mit.edu:1721.1/548742022-01-13T07:54:36Zcom_1721.1_7582com_1721.1_7581col_1721.1_131022

Franz S. Hover. Greytak, Matthew B. (Matthew Bardeen) Massachusetts Institute of Technology. Dept. of Mechanical Engineering. Massachusetts Institute of Technology. Department of Mechanical Engineering 2010-05-25T19:22:34Z 2010-05-25T19:22:34Z 2009 2009 http://hdl.handle.net/1721.1/54874 612391261 Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Mechanical Engineering, 2009. This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections. Cataloged from student-submitted PDF version of thesis. Includes bibliographical references (p. 269-275). Robust motion planning algorithms for mobile robots consider stochasticity in the dynamic model of the vehicle and the environment. A practical robust planning approach balances the duration of the motion plan with the probability of colliding with obstacles. This thesis develops fast analytic algorithms for predicting the collision probability due to model uncertainty and random disturbances in the environment for a planar holonomic vehicle such as a marine surface vessel. These predictions lead to a robust motion planning algorithm that nds the optimal motion plan quickly and efficiently. By incorporating model learning into the predictions, the integrated algorithm exhibits emergent active learning strategies to autonomously acquire the model data needed to safely and eectively complete the mission. The motion planner constructs plans through a known environment by concatenating maneuvers based upon speed controller setpoints. A model-based feedforward/ feedback controller is used to track the resulting reference trajectory, and the model parameters are learned online with a least squares regression algorithm. The path-following performance of the vehicle depends on the effects of unknown environmental disturbances and modeling error. The convergence rate of the parameter estimates depends on the motion plan, as different plans excite different modes of the system. (cont.) By predicting how the collision probability is affected by the parameter covariance evolution, the motion planner automatically incorporates active learning strategies into the motion plans. In particular, the vehicle will practice maneuvers in the open regions of the configuration space before using them in the constrained regions to ensure that the collision risk due to modeling error is low. High-level feedback across missions allows the system to recognize configuration changes and quickly learn new model parameters as necessary. Simulations and experimental results using an autonomous marine surface vessel are presented. by Matthew Greytak. Ph.D. 275 p. eng Massachusetts Institute of Technology M.I.T. theses are protected by copyright. They may be viewed from this source for any purpose, but reproduction or distribution in any format is prohibited without written permission. See provided URL for inquiries about permission. http://dspace.mit.edu/handle/1721.1/7582 Mechanical Engineering. Integrated motion planning and model learning for mobile robots with application to marine vehicles Thesis application/pdf false Publication unknown unknown <Publication xmlns="https://www.openaire.eu/cerif-profile/1.1/" id="69ad847f-0fa8-4a83-b7b0-ccfb074106cf"> <Type xmlns="https://www.openaire.eu/cerif-profile/vocab/COAR_Publication_Types">http://purl.org/coar/resource_type/c_1843</Type> <Language>eng</Language> <Title>Integrated motion planning and model learning for mobile robots with application to marine vehicles</Title> <PublishedIn> <Publication> </Publication> </PublishedIn> <PublicationDate>2009</PublicationDate> <Authors> <Author> <DisplayName>Greytak, Matthew B. (Matthew Bardeen)</DisplayName> <Affiliation> <OrgUnit> </OrgUnit> </Affiliation> </Author> </Authors> <Editors> </Editors> <Publishers> <Publisher> <DisplayName>Massachusetts Institute of Technology</DisplayName> <OrgUnit /> </Publisher> </Publishers> <License>http://dspace.mit.edu/handle/1721.1/7582</License> <Keyword>Mechanical Engineering.</Keyword> <Abstract>Robust motion planning algorithms for mobile robots consider stochasticity in the dynamic model of the vehicle and the environment. A practical robust planning approach balances the duration of the motion plan with the probability of colliding with obstacles. This thesis develops fast analytic algorithms for predicting the collision probability due to model uncertainty and random disturbances in the environment for a planar holonomic vehicle such as a marine surface vessel. These predictions lead to a robust motion planning algorithm that nds the optimal motion plan quickly and efficiently. By incorporating model learning into the predictions, the integrated algorithm exhibits emergent active learning strategies to autonomously acquire the model data needed to safely and eectively complete the mission. The motion planner constructs plans through a known environment by concatenating maneuvers based upon speed controller setpoints. A model-based feedforward/ feedback controller is used to track the resulting reference trajectory, and the model parameters are learned online with a least squares regression algorithm. The path-following performance of the vehicle depends on the effects of unknown environmental disturbances and modeling error. The convergence rate of the parameter estimates depends on the motion plan, as different plans excite different modes of the system.</Abstract> <Abstract>(cont.) By predicting how the collision probability is affected by the parameter covariance evolution, the motion planner automatically incorporates active learning strategies into the motion plans. In particular, the vehicle will practice maneuvers in the open regions of the configuration space before using them in the constrained regions to ensure that the collision risk due to modeling error is low. High-level feedback across missions allows the system to recognize configuration changes and quickly learn new model parameters as necessary. Simulations and experimental results using an autonomous marine surface vessel are presented.</Abstract> <Access xmlns="http://purl.org/coar/access_right" > </Access> </Publication>