Autonomous harvesting may provide a viable solution to mounting labor pressures in the United States' strawberry industry. However, due to bottlenecks in machine perception and economic viability, a profitable and commercially adopted strawberry harvesting system remains elusive. In this research, we explore the feasibility of using deep reinforcement learning to overcome these bottlenecks and develop a practical algorithm to address the sub-objective of viewpoint optimization, or the development of a control policy to direct a camera to favorable vantage points for autonomous harvesting. We evaluate the algorithm's performance in a custom, open-source simulated environment and observe affirmative results. Our trained agent yields 8.7 times higher returns than random actions and 8.8 percent faster exploration than our best baseline policy, which uses visual servoing. Visual investigation shows the agent is able to fixate on favorable viewpoints, despite having no explicit means to propagate information through time. Overall, we conclude that deep reinforcement learning is a promising area of research to advance the state of the art in autonomous strawberry harvesting.
","abstract_html":"<p>Autonomous harvesting may provide a viable solution to mounting labor pressures in the United States' strawberry industry. However, due to bottlenecks in machine perception and economic viability, a profitable and commercially adopted strawberry harvesting system remains elusive. In this research, we explore the feasibility of using deep reinforcement learning to overcome these bottlenecks and develop a practical algorithm to address the sub-objective of viewpoint optimization, or the development of a control policy to direct a camera to favorable vantage points for autonomous harvesting. We evaluate the algorithm's performance in a custom, open-source simulated environment and observe affirmative results. Our trained agent yields 8.7 times higher returns than random actions and 8.8 percent faster exploration than our best baseline policy, which uses visual servoing. Visual investigation shows the agent is able to fixate on favorable viewpoints, despite having no explicit means to propagate information through time. Overall, we conclude that deep reinforcement learning is a promising area of research to advance the state of the art in autonomous strawberry harvesting.</p>","abstract_has_math":false,"creators":["Sather, Jonathon J"],"institution":null,"degree_name":"MS in Mechanical Engineering","degree_level":null,"degree_discipline":"Mechanical Engineering","degree_department":null,"school":null,"contributors":["John Ridgely"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2019,"date_issued":"2019-06-01T07:00:00Z","date_published":"2019-06-01T07:00:00Z","updated_at":"2026-07-24T01:32:13Z","subjects":["reinforcement learning","deep learning","autonomous","autonomous harvesting","automated harvesting","Controls and Control Theory"],"languages":[],"rights":[],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier","label":"Identifier","values":["10.15368/theses.2019.63"],"render_values":[{"text":"10.15368/theses.2019.63","href":"https://doi.org/10.15368/theses.2019.63","code":true}]}]},"links":{"outbound_url":"https://digitalcommons.calpoly.edu/theses/2008","outbound_label":"Repository record","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["John Ridgely"]},{"key":"dc:creator","label":"Author","values":["Sather, Jonathon J"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.available","label":"Dc Date Available","values":["2019-06-22T07:00:00Z"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Mechanical Engineering"]},{"key":"thesis:degree_name","label":"Degree Name","values":["MS in Mechanical Engineering"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["reinforcement learning","deep learning","autonomous","autonomous harvesting","automated harvesting","Controls and Control Theory"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://digitalcommons.calpoly.edu/theses/2008","10.15368/theses.2019.63"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Autonomous harvesting may provide a viable solution to mounting labor pressures in the United States' strawberry industry. However, due to bottlenecks in machine perception and economic viability, a profitable and commercially adopted strawberry harvesting system remains elusive. In this research, we explore the feasibility of using deep reinforcement learning to overcome these bottlenecks and develop a practical algorithm to address the sub-objective of viewpoint optimization, or the development of a control policy to direct a camera to favorable vantage points for autonomous harvesting. We evaluate the algorithm's performance in a custom, open-source simulated environment and observe affirmative results. Our trained agent yields 8.7 times higher returns than random actions and 8.8 percent faster exploration than our best baseline policy, which uses visual servoing. Visual investigation shows the agent is able to fixate on favorable viewpoints, despite having no explicit means to propagate information through time. Overall, we conclude that deep reinforcement learning is a promising area of research to advance the state of the art in autonomous strawberry harvesting.
"]},{"key":"dc:title","label":"Title","values":["Viewpoint Optimization for Autonomous Strawberry Harvesting with Deep Reinforcement Learning"]}]}],"canonical_facts":{"dc:contributor":["John Ridgely"],"dc:creator":["Sather, Jonathon J"],"dc:date.available":["2019-06-22T07:00:00Z"],"dc:description.abstract":["Autonomous harvesting may provide a viable solution to mounting labor pressures in the United States' strawberry industry. However, due to bottlenecks in machine perception and economic viability, a profitable and commercially adopted strawberry harvesting system remains elusive. In this research, we explore the feasibility of using deep reinforcement learning to overcome these bottlenecks and develop a practical algorithm to address the sub-objective of viewpoint optimization, or the development of a control policy to direct a camera to favorable vantage points for autonomous harvesting. We evaluate the algorithm's performance in a custom, open-source simulated environment and observe affirmative results. Our trained agent yields 8.7 times higher returns than random actions and 8.8 percent faster exploration than our best baseline policy, which uses visual servoing. Visual investigation shows the agent is able to fixate on favorable viewpoints, despite having no explicit means to propagate information through time. Overall, we conclude that deep reinforcement learning is a promising area of research to advance the state of the art in autonomous strawberry harvesting.
"],"dc:identifier":["https://digitalcommons.calpoly.edu/theses/2008","10.15368/theses.2019.63"],"dc:subject":["reinforcement learning","deep learning","autonomous","autonomous harvesting","automated harvesting","Controls and Control Theory"],"dc:title":["Viewpoint Optimization for Autonomous Strawberry Harvesting with Deep Reinforcement Learning"],"thesis:degree_discipline":["Mechanical Engineering"],"thesis:degree_name":["MS in Mechanical Engineering"]},"updated_at":"2026-07-24T01:32:13Z"}