RBE 594 Robotics Capstone Final Presentations

Team Blue: Automatic Blueberry Harvester

Aaron Erickson, Walt Gallati, Oleg Russu

Blueberries, the world’s second-most produced berry crop, relies on manual labor for harvesting despite an increasingly scarce and expensive labor pool. We propose an automated harvesting mechanism that performs at similar speeds to human laborers without the produce damage rates of traditional mechanized approaches. The industry needs a system that can harvest 12 lbs of blueberries per hour with 85% of blueberries remaining undamaged. This approach utilizes a compliant rake-based system mounted on a 6-degree-of-freedom robotic arm. Tunable compliance is provided by a pneumatic tine rigidizing system, reducing damage to the berries and stems. Object detection uses machine learning, accurately identifying blueberries and supplying their locations to the path planner for targeted picking. The system is being validated using a combination of physical testing using our calibrated test stand and simulated validation. Test stand results indicate harvester is capable of easily stripping ripe berries from branches. Simulink damage simulation and ROS operational simulation results are pending. If the design goals of the system are met the blueberry harvesting industry could see an uptick in scalable automatic harvesting systems like the one our team has investigated.

Keywords: Agriculture, Controls, Pneumatics, Vision, Automation 

Autonomous Service Robot for Table Clearing in Restaurant Environments 

Chris Dwight, Howard Ho, Tom Mulroy, Harsh Verma 

Labor shortages impact the long-term sustainability of restaurants of all sizes. Our project demonstrates the design and simulation results of an autonomous mobile robot capable of clearing tables in a restaurant setting. The robot navigates to tables, identifies and collects dirty items using a robotic arm, and transports them to a designated kitchen area. Key components include SLAM for localization, autonomous navigation, computer vision for object recognition, and arm motion planning using both traditional and reinforcement learning approaches. The end result is a robot that can accurately and reliably fulfill the role of a human busser, allowing restaurants to maintain consistent operations with reduced staffing. 

Automated Robotic Waste Sorting System for Enhanced Recycling Efficiency

Gabriel Demanche, Gabrielle Vanner, Nate Dixon, Will Yingling

As global temperature rises and local climates shift and evolve, citizens are becoming more aware of the impacts climate change has on their daily lives. A large contributor to climate change is waste production, which causes greenhouse gas emissions, pollution, and land degradation. Recycling rates across the United States are extremely low, with some estimates below 32%, due to high costs and challenges in the recycling process. Waste sorting is tedious and potentially hazardous to human workers, as they come into contact with biological contaminants, sharp objects, and chemicals. This project focuses on the development of an automated waste sorting system that uses artificial intelligence to detect contaminants and classify recyclable material by type. The waste is sorted by a robotic manipulator, allowing for higher accuracy and safety for the workers. The identified material is placed into bins to be processed and cleaned, streamlining the recycling process and improving overall efficiency.

No RSVP necessary! Simply join via the zoom link: Zoom meeting has ended.