{"id":{"repo_id":"msu","oai_identifier":"oai:d.lib.msu.edu:etd_53303"},"canonical_url":"https://search.ndltd.org/etd/msu/oai:d.lib.msu.edu:etd_53303","repository":{"repo_id":"msu","name":"Michigan State University","base_url":"https://d.lib.msu.edu/oai"},"display":{"title":"Learning-enabled control design for safety-critical systems operating in cluttered environments","abstract":"Thesis (Ph.D.)--Michigan State University. Electrical and Computer Engineering - Doctor of Philosophy, 2026","abstract_html":"Thesis (Ph.D.)--Michigan State University. Electrical and Computer Engineering - Doctor of Philosophy, 2026","abstract_has_math":false,"creators":["Khaledi, Marjan"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":["Kiumarsi, Bahare","Tan, Xiaobo","Radha, Hayder","Modares, Hamidreza"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026","date_published":"2026","updated_at":"2026-11-18T18:59:47Z","subjects":["Electrical engineering","Robotics"],"languages":["eng"],"rights":["In Copyright"],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier","label":"Identifier","values":["etd:53303","local:Khaledi_grad.msu_0128D_22356"],"render_values":[{"text":"etd:53303","href":null,"code":true},{"text":"local:Khaledi_grad.msu_0128D_22356","href":null,"code":true}]}],"person_identifier_entries":[]},"links":{"outbound_url":"https://doi.org/doi:10.25335/d00m-bv67","outbound_label":"DOI","outbound_source":"dc:identifier","record_url":"https://doi.org/doi:10.25335/d00m-bv67","record_label":"Repository record","record_kind":"doi","record_source":"dc:identifier","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://d.lib.msu.edu/oai?verb=GetRecord&metadataPrefix=oai_dc&identifier=oai%3Ad.lib.msu.edu%3Aetd_53303","prefix":"oai_dc"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Kiumarsi, Bahare","Tan, Xiaobo","Radha, Hayder","Modares, Hamidreza"]},{"key":"dc:creator","label":"Author","values":["Khaledi, Marjan"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2026"]},{"key":"dc:relation","label":"Dc Relation","values":["Electronic Theses & Dissertations"]},{"key":"dc:type","label":"Dc Type","values":["text","Electronic dissertations"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Electrical engineering","Robotics"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["eng"]},{"key":"dc:rights","label":"Dc Rights","values":["In Copyright"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["etd:53303","local:Khaledi_grad.msu_0128D_22356","https://doi.org/doi:10.25335/d00m-bv67"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Thesis (Ph.D.)--Michigan State University. Electrical and Computer Engineering - Doctor of Philosophy, 2026","Autonomous systems are increasingly expected to operate safely in unknown, constrained, and dynamically changing environments. However, guaranteeing safety while maintaining performance and computational efficiency remains challenging, particularly when accurate system models are unavailable. This dissertation develops safety-critical data-driven control and predictive control methods using barrier function-based certificates, optimization, and behavioral system representations to address these challenges.First, a safety-certified data-driven predictive control framework is developed for linear systems by combining behavioral systems theory with control barrier certificates. The proposed formulation uses measured input-state trajectories instead of explicit system matrices and constructs a data-driven maximal safe terminal invariant set. This design improves feasibility, reduces the required prediction horizon, and guarantees recursive feasibility and stability.Second, safety-certified model predictive control frameworks based on dynamic control barrier functions are developed for robotic navigation in environments containing both static and moving obstacles. Bounded relaxation mechanisms are introduced to improve feasibility while preserving safety guarantees. Furthermore, to reduce computational complexity, a soft-minimum composite dynamic control barrier function is proposed, allowing multiple obstacle-avoidance requirements to be represented by a single smooth safety constraint at each prediction step. Numerical and Gazebo-based ROS2 simulations, along with experimental results, demonstrate safe and efficient navigation in cluttered dynamic environments. Next, a safety-critical data-driven predictive control framework is developed for nonlinear robotic systems operating in cluttered environments with multiple static obstacles. Unlike the previous data-driven predictive control framework for linear systems, which uses control barrier certificates and a data-driven terminal invariant set, this approach integrates discrete-time control barrier function constraints directly into a Hankel-matrix-based behavioral predictive control formulation. Future trajectories are generated from measured input-state data through a local behavioral approximation of the nonlinear system, while regularization and slack variables account for prediction mismatch caused by nonlinear dynamics and finite data. A bounded relaxation mechanism is incorporated into the barrier constraints to improve feasibility in constrained environments. Recursive feasibility, practical safety, and practical stability are established. Gazebo-based ROS2 simulations and experimental results further demonstrate the effectiveness of the proposed approach in multiple-obstacle avoidance scenarios. This dissertation also develops a computationally efficient data-driven safe optimal control framework for continuous-time systems with unknown dynamics. This approach learns safe and optimal controllers directly from data and combines them through a control-merging mechanism that prioritizes safety near the boundary of the safe set and optimality away from safety-critical regions. The final part of this dissertation extends data-driven safety-critical control to distributed multi-agent systems by developing a direct data-driven control barrier function-based quadratic programming framework that enables agents to satisfy safety constraints using locally available information.Overall, the results of this dissertation show that safety-critical controllers can be designed directly from data, incorporated into predictive control frameworks, and extended to robotic and multi-agent systems while preserving formal safety guarantees. The proposed methods reduce dependence on accurate models, improve feasibility in constrained environments, and provide effective mechanisms for balancing safety, performance, and computational efficiency in autonomous systems.","Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).","Includes bibliographical references."]},{"key":"dc:format","label":"Dc Format","values":["application/pdf","1 online resource (178 p.) : ill."]},{"key":"dc:title","label":"Title","values":["Learning-enabled control design for safety-critical systems operating in cluttered environments"]}]}],"canonical_facts":{"dc:contributor":["Kiumarsi, Bahare","Tan, Xiaobo","Radha, Hayder","Modares, Hamidreza"],"dc:creator":["Khaledi, Marjan"],"dc:date":["2026"],"dc:description":["Thesis (Ph.D.)--Michigan State University. Electrical and Computer Engineering - Doctor of Philosophy, 2026","Autonomous systems are increasingly expected to operate safely in unknown, constrained, and dynamically changing environments. However, guaranteeing safety while maintaining performance and computational efficiency remains challenging, particularly when accurate system models are unavailable. This dissertation develops safety-critical data-driven control and predictive control methods using barrier function-based certificates, optimization, and behavioral system representations to address these challenges.First, a safety-certified data-driven predictive control framework is developed for linear systems by combining behavioral systems theory with control barrier certificates. The proposed formulation uses measured input-state trajectories instead of explicit system matrices and constructs a data-driven maximal safe terminal invariant set. This design improves feasibility, reduces the required prediction horizon, and guarantees recursive feasibility and stability.Second, safety-certified model predictive control frameworks based on dynamic control barrier functions are developed for robotic navigation in environments containing both static and moving obstacles. Bounded relaxation mechanisms are introduced to improve feasibility while preserving safety guarantees. Furthermore, to reduce computational complexity, a soft-minimum composite dynamic control barrier function is proposed, allowing multiple obstacle-avoidance requirements to be represented by a single smooth safety constraint at each prediction step. Numerical and Gazebo-based ROS2 simulations, along with experimental results, demonstrate safe and efficient navigation in cluttered dynamic environments. Next, a safety-critical data-driven predictive control framework is developed for nonlinear robotic systems operating in cluttered environments with multiple static obstacles. Unlike the previous data-driven predictive control framework for linear systems, which uses control barrier certificates and a data-driven terminal invariant set, this approach integrates discrete-time control barrier function constraints directly into a Hankel-matrix-based behavioral predictive control formulation. Future trajectories are generated from measured input-state data through a local behavioral approximation of the nonlinear system, while regularization and slack variables account for prediction mismatch caused by nonlinear dynamics and finite data. A bounded relaxation mechanism is incorporated into the barrier constraints to improve feasibility in constrained environments. Recursive feasibility, practical safety, and practical stability are established. Gazebo-based ROS2 simulations and experimental results further demonstrate the effectiveness of the proposed approach in multiple-obstacle avoidance scenarios. This dissertation also develops a computationally efficient data-driven safe optimal control framework for continuous-time systems with unknown dynamics. This approach learns safe and optimal controllers directly from data and combines them through a control-merging mechanism that prioritizes safety near the boundary of the safe set and optimality away from safety-critical regions. The final part of this dissertation extends data-driven safety-critical control to distributed multi-agent systems by developing a direct data-driven control barrier function-based quadratic programming framework that enables agents to satisfy safety constraints using locally available information.Overall, the results of this dissertation show that safety-critical controllers can be designed directly from data, incorporated into predictive control frameworks, and extended to robotic and multi-agent systems while preserving formal safety guarantees. The proposed methods reduce dependence on accurate models, improve feasibility in constrained environments, and provide effective mechanisms for balancing safety, performance, and computational efficiency in autonomous systems.","Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).","Includes bibliographical references."],"dc:format":["application/pdf","1 online resource (178 p.) : ill."],"dc:identifier":["etd:53303","local:Khaledi_grad.msu_0128D_22356","https://doi.org/doi:10.25335/d00m-bv67"],"dc:language":["eng"],"dc:relation":["Electronic Theses & Dissertations"],"dc:rights":["In Copyright"],"dc:subject":["Electrical engineering","Robotics"],"dc:title":["Learning-enabled control design for safety-critical systems operating in cluttered environments"],"dc:type":["text","Electronic dissertations"]},"updated_at":"2026-11-18T18:59:47Z"}