Location: Sugarbeet and Bean Research
Project Number: 5050-43640-003-012-I
Project Type: Interagency Reimbursable Agreement
Start Date: Sep 15, 2023
End Date: Sep 14, 2027
Objective:
1. Design and construct efficient, dexterous multi-arm robotic harvesting modules, coupled with AI algorithms and optimum systems control and planning schemes, to enable picking apples grown in high-density tree orchards.
2. Develop a low-cost and compact machine vision-based in-field pre-sorting system to segregate low-quality or inferior fruit and record quality information for harvested apples.
3. Design and construct a new Automated and Integrated Mobile System (AIMS) with fully integrated robotic harvesting and in-field pre-sorting modules, along with automated fruit and bin handling functions, to enable selective or full harvesting, pre-sorting, and quality tracking of apples in the orchard.
4. Evaluate and demonstrate the AIMS in diverse commercial orchard systems and with different operational strategies in Michigan, Pennsylvania, and Washington.
5. Establish open-access repositories of labeled image data collected by multi-modal sensing systems under diverse orchard conditions as well as a performance benchmark suite of state-of-the-art AI models for fruit detection and quality grading.
6. Conduct cost benefit analyses of automated harvesting and in-field sorting technologies and the impact of different models of technology adoption on the U.S. apple industry and the future labor force.
Approach:
Multi-arm robotic harvesting modules will be designed and constructed, with augmented algorithms for localizing fruit, foliage and branches and for planning and coordinating multiple arms to enhance the overall picking performance of the robots. A new low-cost and compact in-field pre-sorting system will be designed and constructed, incorporating with a new AI algorithm for full-scale inspection and quality recording of harvested apples for color, size and defects. We will also build an artificial lighting and controlling configuration to allow the robot to work in the orchard during day and at night. An autonomous mobile harvest platform will be designed and constructed for integration with the multi-arm robotic modules and the new in-field pre-sorting system. In addition, automated bin filling and handling functions will be incorporated with the mobile platform. Field tests and demonstrations of the robotic harvesting modules and AIMS will be conducted in Michigan, Pennsylvania and Washington. After appropriate IP protection, the results for the AIMS will be disseminated to growers, extension educators and industry partners through presentations, written publications, and webinars. Economic/labor analyses of the developed technologies, different technology adoption models and different picking strategies (selective vs full picking) will be conducted to determine their impact on the apple industry and the future labor force.
