{"id":{"repo_id":"mtu","oai_identifier":"oai:digitalcommons.mtu.edu:etdr-2788"},"canonical_url":"https://search.ndltd.org/etd/mtu/oai:digitalcommons.mtu.edu:etdr-2788","repository":{"repo_id":"mtu","name":"Michigan Technological University","base_url":"https://digitalcommons.mtu.edu/do/oai/"},"display":{"title":"BENCHMARKING MODEL PREDICTIVE CONTROL AND REINFORCEMENT LEARNING FOR LEGGED ROBOT LOCOMOTION","abstract":"This research delves into the realm of quadrupedal robotics, focusing on the comparative analysis of Model Predictive Control (MPC) and Reinforcement Learning (RL) as predominant control strategies. Through the comprehensive dataset compiled and the insights derived from this analysis, this research aims to serve as a valuable resource for the legged robotics community, guiding researchers and practitioners in the selection and implementation of control strategies. The ultimate goal is to contribute to the advancement of legged robot capabilities and facilitate their successful deployment in real-world applications. In this study, we employ the Unitree Go1 quadrupedal robot as a testbed, subjecting it to a variety of conditions including different terrains and external perturbations to assess the performance of MPC and RL controllers. Our findings reveal that RL exhibits superior force rejection in scenarios involving external forces, albeit relying heavily on torque in a single joint, while MPC provides a balanced torque distribution across all joints. In stumbling scenarios, MPC outperforms RL in recovery time, although both controllers face challenges when the robot falls into a failure state. Furthermore, the generalization capabilities of RL are evaluated across different terrains, demonstrating a performance drop in slippery conditions and uneven terrains compared to flat frictional surfaces. The temporal demands of RL, encompassing optimization and training phases, are contrasted with the real-time operation and parameter flexibility of MPC.","abstract_html":"This research delves into the realm of quadrupedal robotics, focusing on the comparative analysis of Model Predictive Control (MPC) and Reinforcement Learning (RL) as predominant control strategies. Through the comprehensive dataset compiled and the insights derived from this analysis, this research aims to serve as a valuable resource for the legged robotics community, guiding researchers and practitioners in the selection and implementation of control strategies. The ultimate goal is to contribute to the advancement of legged robot capabilities and facilitate their successful deployment in real-world applications. In this study, we employ the Unitree Go1 quadrupedal robot as a testbed, subjecting it to a variety of conditions including different terrains and external perturbations to assess the performance of MPC and RL controllers. Our findings reveal that RL exhibits superior force rejection in scenarios involving external forces, albeit relying heavily on torque in a single joint, while MPC provides a balanced torque distribution across all joints. In stumbling scenarios, MPC outperforms RL in recovery time, although both controllers face challenges when the robot falls into a failure state. Furthermore, the generalization capabilities of RL are evaluated across different terrains, demonstrating a performance drop in slippery conditions and uneven terrains compared to flat frictional surfaces. The temporal demands of RL, encompassing optimization and training phases, are contrasted with the real-time operation and parameter flexibility of MPC.","abstract_has_math":false,"creators":["Akki, Shivayogi"],"institution":null,"degree_name":"Master of Science in Mechatronics (MS)","degree_level":"Open Access Master's Report","degree_discipline":"Department of Electrical and Computer Engineering","degree_department":null,"school":null,"contributors":["Tan Chen"],"advisors":[],"committee_chairs":[],"committee_members":["Vinh Nguyen","Tao Liu"],"year":null,"date_issued":"","date_published":null,"updated_at":"2026-11-19T08:18:53Z","subjects":["Legged Robotics","Reinforcement Learning","Model Predictive Control","Locomotion","Robotics"],"languages":[],"rights":[],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier","label":"Identifier","values":["https://digitalcommons.mtu.edu/etdr/1677"],"render_values":[{"text":"https://digitalcommons.mtu.edu/etdr/1677","href":"https://digitalcommons.mtu.edu/etdr/1677","code":true}]}],"person_identifier_entries":[]},"links":{"outbound_url":"https://doi.org/10.37099/mtu.dc.etdr/1677","outbound_label":"DOI","outbound_source":"dc:identifier","record_url":"https://doi.org/10.37099/mtu.dc.etdr/1677","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://digitalcommons.mtu.edu/do/oai/?verb=GetRecord&metadataPrefix=oai_etdms&identifier=oai%3Adigitalcommons.mtu.edu%3Aetdr-2788","prefix":"oai_etdms"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Tan Chen"]},{"key":"dc:contributor.committeemember","label":"Committee Member","values":["Vinh Nguyen","Tao Liu"]},{"key":"dc:creator","label":"Author","values":["Akki, Shivayogi"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.created","label":"Dc Date Created","values":["2023-01-01T08:00:00Z"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Department of Electrical and Computer Engineering"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Open Access Master's Report"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Master of Science in Mechatronics (MS)"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Legged Robotics","Reinforcement Learning","Model Predictive Control","Locomotion","Robotics"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["10.37099/mtu.dc.etdr/1677","https://digitalcommons.mtu.edu/etdr/1677"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["This research delves into the realm of quadrupedal robotics, focusing on the comparative analysis of Model Predictive Control (MPC) and Reinforcement Learning (RL) as predominant control strategies. Through the comprehensive dataset compiled and the insights derived from this analysis, this research aims to serve as a valuable resource for the legged robotics community, guiding researchers and practitioners in the selection and implementation of control strategies. The ultimate goal is to contribute to the advancement of legged robot capabilities and facilitate their successful deployment in real-world applications. In this study, we employ the Unitree Go1 quadrupedal robot as a testbed, subjecting it to a variety of conditions including different terrains and external perturbations to assess the performance of MPC and RL controllers. Our findings reveal that RL exhibits superior force rejection in scenarios involving external forces, albeit relying heavily on torque in a single joint, while MPC provides a balanced torque distribution across all joints. In stumbling scenarios, MPC outperforms RL in recovery time, although both controllers face challenges when the robot falls into a failure state. Furthermore, the generalization capabilities of RL are evaluated across different terrains, demonstrating a performance drop in slippery conditions and uneven terrains compared to flat frictional surfaces. The temporal demands of RL, encompassing optimization and training phases, are contrasted with the real-time operation and parameter flexibility of MPC."]},{"key":"dc:title","label":"Title","values":["BENCHMARKING MODEL PREDICTIVE CONTROL AND REINFORCEMENT LEARNING FOR LEGGED ROBOT LOCOMOTION"]}]}],"canonical_facts":{"dc:contributor":["Tan Chen"],"dc:contributor.committeemember":["Vinh Nguyen","Tao Liu"],"dc:creator":["Akki, Shivayogi"],"dc:date.created":["2023-01-01T08:00:00Z"],"dc:description.abstract":["This research delves into the realm of quadrupedal robotics, focusing on the comparative analysis of Model Predictive Control (MPC) and Reinforcement Learning (RL) as predominant control strategies. Through the comprehensive dataset compiled and the insights derived from this analysis, this research aims to serve as a valuable resource for the legged robotics community, guiding researchers and practitioners in the selection and implementation of control strategies. The ultimate goal is to contribute to the advancement of legged robot capabilities and facilitate their successful deployment in real-world applications. In this study, we employ the Unitree Go1 quadrupedal robot as a testbed, subjecting it to a variety of conditions including different terrains and external perturbations to assess the performance of MPC and RL controllers. Our findings reveal that RL exhibits superior force rejection in scenarios involving external forces, albeit relying heavily on torque in a single joint, while MPC provides a balanced torque distribution across all joints. In stumbling scenarios, MPC outperforms RL in recovery time, although both controllers face challenges when the robot falls into a failure state. Furthermore, the generalization capabilities of RL are evaluated across different terrains, demonstrating a performance drop in slippery conditions and uneven terrains compared to flat frictional surfaces. The temporal demands of RL, encompassing optimization and training phases, are contrasted with the real-time operation and parameter flexibility of MPC."],"dc:identifier":["10.37099/mtu.dc.etdr/1677","https://digitalcommons.mtu.edu/etdr/1677"],"dc:subject":["Legged Robotics","Reinforcement Learning","Model Predictive Control","Locomotion","Robotics"],"dc:title":["BENCHMARKING MODEL PREDICTIVE CONTROL AND REINFORCEMENT LEARNING FOR LEGGED ROBOT LOCOMOTION"],"thesis:degree_discipline":["Department of Electrical and Computer Engineering"],"thesis:degree_level":["Open Access Master's Report"],"thesis:degree_name":["Master of Science in Mechatronics (MS)"]},"updated_at":"2026-11-19T08:18:53Z"}