Mark Presten

EECS Department, University of California, Berkeley

Technical Report No. UCB/EECS-2022-95

May 13, 2022

http://www2.eecs.berkeley.edu/Pubs/TechRpts/2022/EECS-2022-95.pdf

Polyculture farming - where multiple crop species are grown simultaneously - has potential to reduce pesticide and water usage while improving the utilization of soil nutrients. However, it is much harder to automate polyculture than monoculture. This report presents two contributions to the research and development of polyculture farming, with the first being AlphaGardenSim: a fast, first order, open-access polyculture farming simulator with single plant growth and irrigation models tuned using real world measurements. AlphaGardenSim can be used for policy learning as it simulates inter-plant dynamics, including light and water competition between plants in close proximity and approximates growth in a real greenhouse garden at 25,000X the speed of natural growth. We discuss the development of the simulator, the models used for growth, light, and irrigation, real-to-sim model tuning methods, policies trained in simulator, and metrics and results for simulated garden cycles.

The latter half of this report presents AlphaGarden: an automated system for pruning and irrigating living plants in a physical testbed that uses policies in AlphaGardenSim to decide real-time actions. This system utilizes novel hardware and algorithms for automated pruning. Using an overhead camera to collect data from a physical garden testbed, the autonomous system utilizes a learned Plant Phenotyping convolutional neural network and a Bounding Disk Tracking algorithm to evaluate the individual plant distribution and estimate the state of the garden each day. From this garden state, AlphaGardenSim selects plants to prune. A trained neural network detects and targets specific prune points on the plant. Two custom-designed pruning tools, compatible with a FarmBot gantry system, are experimentally evaluated and execute autonomous cuts through controlled algorithms. We show results for four 60-day garden cycles. Results suggest the system can autonomously achieve 0.94 normalized plant diversity with pruning shears while maintaining an average canopy coverage of 0.84 by the end of the cycles. In ongoing work, we optimize water usage and also compare the AlphaGarden system to a human gardener.

Advisors: Ken Goldberg

---

BibTeX citation:

@mastersthesis{Presten:EECS-2022-95,
    Author= {Presten, Mark},
    Editor= {Goldberg, Ken},
    Title= {Design and Implementation of Physical Experiments for Evaluation of the AlphaGarden: an Autonomous Polyculture Garden},
    School= {EECS Department, University of California, Berkeley},
    Year= {2022},
    Month= {May},
    Url= {http://www2.eecs.berkeley.edu/Pubs/TechRpts/2022/EECS-2022-95.html},
    Number= {UCB/EECS-2022-95},
    Abstract= {Polyculture farming - where multiple crop species are grown simultaneously - has potential to reduce pesticide and water usage while improving the utilization of soil nutrients. However, it is much harder to automate polyculture than monoculture. This report presents two contributions to the research and development of polyculture farming, with the first being AlphaGardenSim: a fast, first order, open-access polyculture farming simulator with single plant growth and irrigation models tuned using real world measurements. AlphaGardenSim can be used for policy learning as it simulates inter-plant dynamics, including light and water competition between plants in close proximity and approximates growth in a real greenhouse garden at 25,000X the speed of natural growth. We discuss the development of the simulator, the models used for growth, light, and irrigation, real-to-sim model tuning methods, policies trained in simulator, and metrics and results for simulated garden cycles. 

The latter half of this report presents AlphaGarden: an automated system for pruning and irrigating living plants in a physical testbed that uses policies in AlphaGardenSim to decide real-time actions. This system utilizes novel hardware and algorithms for automated pruning. Using an overhead camera to collect data from a physical garden testbed, the autonomous system utilizes a learned Plant Phenotyping convolutional neural network and a Bounding Disk Tracking algorithm to evaluate the individual plant distribution and estimate the state of the garden each day. From this garden state, AlphaGardenSim selects plants to prune. A trained neural network detects and targets specific prune points on the plant. Two custom-designed pruning tools, compatible with a FarmBot gantry system, are experimentally evaluated and execute autonomous cuts through controlled algorithms. We show results for four 60-day garden cycles. Results suggest the system can autonomously achieve 0.94 normalized plant diversity with pruning shears while maintaining an average canopy coverage of 0.84 by the end of the cycles. In ongoing work, we optimize water usage and also compare the AlphaGarden system to a human gardener.},
}

EndNote citation:

%0 Thesis
%A Presten, Mark 
%E Goldberg, Ken 
%T Design and Implementation of Physical Experiments for Evaluation of the AlphaGarden: an Autonomous Polyculture Garden
%I EECS Department, University of California, Berkeley
%D 2022
%8 May 13
%@ UCB/EECS-2022-95
%U http://www2.eecs.berkeley.edu/Pubs/TechRpts/2022/EECS-2022-95.html
%F Presten:EECS-2022-95