{"id":{"repo_id":"msu","oai_identifier":"oai:d.lib.msu.edu:etd_52887"},"canonical_url":"https://search.ndltd.org/etd/msu/oai:d.lib.msu.edu:etd_52887","repository":{"repo_id":"msu","name":"Michigan State University","base_url":"https://d.lib.msu.edu/oai"},"display":{"title":"Safe Performance-Aware Control and Motion Planning Using Prior and Posterior Information","abstract":"Thesis (Ph.D.)--Michigan State University. Mechanical Engineering - Doctor of Philosophy, 2025","abstract_html":"Thesis (Ph.D.)--Michigan State University. Mechanical Engineering - Doctor of Philosophy, 2025","abstract_has_math":false,"creators":["Niknejad, Nariman"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":["Modares, Hamidreza HM"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2025,"date_issued":"2025","date_published":"2025","updated_at":"2026-11-18T18:59:31Z","subjects":["Mechanical engineering","Engineering"],"languages":["eng"],"rights":["In Copyright"],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier","label":"Identifier","values":["etd:52887","local:Niknejad_grad.msu_0128D_21878"],"render_values":[{"text":"etd:52887","href":null,"code":true},{"text":"local:Niknejad_grad.msu_0128D_21878","href":null,"code":true}]}],"person_identifier_entries":[]},"links":{"outbound_url":"https://doi.org/doi:10.25335/z04x-yk43","outbound_label":"DOI","outbound_source":"dc:identifier","record_url":"https://doi.org/doi:10.25335/z04x-yk43","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_52887","prefix":"oai_dc"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Modares, Hamidreza HM"]},{"key":"dc:creator","label":"Author","values":["Niknejad, Nariman"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2025"]},{"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":["Mechanical engineering","Engineering"]}]},{"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:52887","local:Niknejad_grad.msu_0128D_21878","https://doi.org/doi:10.25335/z04x-yk43"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Thesis (Ph.D.)--Michigan State University. Mechanical Engineering - Doctor of Philosophy, 2025","This dissertation develops a unified, data-driven framework for safe and performance-aware control and motion planning in uncertain, noisy, and partially modeled systems. It fuses limited or noisy data with physics-based prior knowledge to (i) certify safety with high probability, (ii) synthesize stabilizing and optimal controllers online, and (iii) plan motions through provably safe invariant sets—while maintaining tractability through convex programs such as LMIs and LPs and set-theoretic formulations. The framework extends to perception-based observer design and tube-based MPC schemes, ensuring bounded error under heavy-tailed noise.Motivation. Real-world systems often lack accurate models, and collecting identification-quality data can be unsafe or impractical. Moreover, perception and process noise are frequently biased or heavy-tailed, invalidating Gaussian or certainty-equivalent assumptions, while worst-case robust control can be overly conservative. These challenges motivate a direct, probabilistic, and physics-informed data-driven approach that (a) guarantees safety with limited data, (b) accommodates non-Gaussian disturbances, and (c) leverages prior physics to reduce conservatism and enhance performance.Methodology. Five core principles underlie the framework:(1) Set intersections between physics-informed and data-consistent models to reduce uncertainty; (2) Invariant and contractive sets, such as ellipsoids or zonotopes, for safety certification; (3) Lossy S-lemma based convex reformulations for probabilistic stability and constraint satisfaction; (4) Online refinement of prior sets as data accrue; and (5) Computational pragmatism, emphasizing LMIs, LPs, and offline computation to ensure real-time feasibility.Chapter Contributions:(1) Online reference tracking develops a two-stage learner combining fast identification and memory-augmented control, achieving order T^{2/3} regret against H-infinity and LQR baselines. (2) Data-driven safe tracking integrates learned feedback and feedforward control with a reference governor enforcing lambda-contractivity of an ellipsoidal safe set, ensuring high-probability constraint satisfaction. (3) Physics-informed safe and optimal control enforces LMIs over the intersection of data- and physics-consistent models, yielding less conservative and more feasible designs than robust baselines. (4) DaSP-RRT recasts motion planning as traversing overlapping invariant ellipsoids with precomputed safe sets and gains. (5) DaSP-LLM integrates large language models with data-driven invariant sets, verifying and refining LLM-generated waypoints through safety checks and structured feedback to guarantee feasibility. (6) Online stabilizing control uses ARX and behavior-theoretic models with a strict lossy S-lemma to derive necessary and sufficient online stability conditions directly from noisy input–output data. (7) Chance-constrained covariance control formulates input and output covariance constraints as probabilistic LMIs that adapt online, ensuring constraint satisfaction under unbounded process noise. (8) Robust MPC with perception-based observers employs constrained zonotopes and tube-based MPC with Minkowski infinity norms, replacing Gaussian assumptions with a zonotopic observer to bound heavy-tailed perception errors.Validation. The framework is demonstrated on drone hovering, omnidirectional robots, active suspension, battery regulation, and spacecraft docking. Across all applications, the methods achieve tighter safety margins, improved feasibility, and lower control cost compared with robust or certainty-equivalent baselines, while satisfying real-time constraints through convex reformulations and precomputation.Takeaway. By unifying physics-based priors with limited and noisy data, certifying safety through invariant sets, and emphasizing convexity and offline computation, this dissertation establishes a scalable methodology for safe, performance-aware autonomy. It provides theoretical and empirical guarantees across the full hierarchy of control and planning—bridging data-driven learning, real-time safety assurance, and physics-consistent optimization under modern sensing and uncertainty conditions.","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 (294 p.) : ill."]},{"key":"dc:title","label":"Title","values":["Safe Performance-Aware Control and Motion Planning Using Prior and Posterior Information"]}]}],"canonical_facts":{"dc:contributor":["Modares, Hamidreza HM"],"dc:creator":["Niknejad, Nariman"],"dc:date":["2025"],"dc:description":["Thesis (Ph.D.)--Michigan State University. Mechanical Engineering - Doctor of Philosophy, 2025","This dissertation develops a unified, data-driven framework for safe and performance-aware control and motion planning in uncertain, noisy, and partially modeled systems. It fuses limited or noisy data with physics-based prior knowledge to (i) certify safety with high probability, (ii) synthesize stabilizing and optimal controllers online, and (iii) plan motions through provably safe invariant sets—while maintaining tractability through convex programs such as LMIs and LPs and set-theoretic formulations. The framework extends to perception-based observer design and tube-based MPC schemes, ensuring bounded error under heavy-tailed noise.Motivation. Real-world systems often lack accurate models, and collecting identification-quality data can be unsafe or impractical. Moreover, perception and process noise are frequently biased or heavy-tailed, invalidating Gaussian or certainty-equivalent assumptions, while worst-case robust control can be overly conservative. These challenges motivate a direct, probabilistic, and physics-informed data-driven approach that (a) guarantees safety with limited data, (b) accommodates non-Gaussian disturbances, and (c) leverages prior physics to reduce conservatism and enhance performance.Methodology. Five core principles underlie the framework:(1) Set intersections between physics-informed and data-consistent models to reduce uncertainty; (2) Invariant and contractive sets, such as ellipsoids or zonotopes, for safety certification; (3) Lossy S-lemma based convex reformulations for probabilistic stability and constraint satisfaction; (4) Online refinement of prior sets as data accrue; and (5) Computational pragmatism, emphasizing LMIs, LPs, and offline computation to ensure real-time feasibility.Chapter Contributions:(1) Online reference tracking develops a two-stage learner combining fast identification and memory-augmented control, achieving order T^{2/3} regret against H-infinity and LQR baselines. (2) Data-driven safe tracking integrates learned feedback and feedforward control with a reference governor enforcing lambda-contractivity of an ellipsoidal safe set, ensuring high-probability constraint satisfaction. (3) Physics-informed safe and optimal control enforces LMIs over the intersection of data- and physics-consistent models, yielding less conservative and more feasible designs than robust baselines. (4) DaSP-RRT recasts motion planning as traversing overlapping invariant ellipsoids with precomputed safe sets and gains. (5) DaSP-LLM integrates large language models with data-driven invariant sets, verifying and refining LLM-generated waypoints through safety checks and structured feedback to guarantee feasibility. (6) Online stabilizing control uses ARX and behavior-theoretic models with a strict lossy S-lemma to derive necessary and sufficient online stability conditions directly from noisy input–output data. (7) Chance-constrained covariance control formulates input and output covariance constraints as probabilistic LMIs that adapt online, ensuring constraint satisfaction under unbounded process noise. (8) Robust MPC with perception-based observers employs constrained zonotopes and tube-based MPC with Minkowski infinity norms, replacing Gaussian assumptions with a zonotopic observer to bound heavy-tailed perception errors.Validation. The framework is demonstrated on drone hovering, omnidirectional robots, active suspension, battery regulation, and spacecraft docking. Across all applications, the methods achieve tighter safety margins, improved feasibility, and lower control cost compared with robust or certainty-equivalent baselines, while satisfying real-time constraints through convex reformulations and precomputation.Takeaway. By unifying physics-based priors with limited and noisy data, certifying safety through invariant sets, and emphasizing convexity and offline computation, this dissertation establishes a scalable methodology for safe, performance-aware autonomy. It provides theoretical and empirical guarantees across the full hierarchy of control and planning—bridging data-driven learning, real-time safety assurance, and physics-consistent optimization under modern sensing and uncertainty conditions.","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 (294 p.) : ill."],"dc:identifier":["etd:52887","local:Niknejad_grad.msu_0128D_21878","https://doi.org/doi:10.25335/z04x-yk43"],"dc:language":["eng"],"dc:relation":["Electronic Theses & Dissertations"],"dc:rights":["In Copyright"],"dc:subject":["Mechanical engineering","Engineering"],"dc:title":["Safe Performance-Aware Control and Motion Planning Using Prior and Posterior Information"],"dc:type":["text","Electronic dissertations"]},"updated_at":"2026-11-18T18:59:31Z"}