New System Helps Maintain Orchard Crop Health and Accurately Estimates Harvests
Outcome/Accomplishment
Understanding plant health is essential for making key farming decisions, like how much fertilizer or pesticide to use and how much produce to expect at harvest. A system of novel systems for keeping track of individual trees and even single fruits in an apple orchard during the growing season has been developed by researchers at the National Science Foundation (NSF)-funded Internet of Things for Precision Agriculture (NSF IoT4Ag) Engineering Research Center (ERC), headquartered at the University of Pennsylvania.
Impact/Benefits
NSF IoT4Ag’s system uses uncrewed aerial vehicles (UAVs) to navigate under foliage canopies, which is essential for tracking changes in crops—including tree crops (Figure 1). The system creates detailed 3D and 4D maps of orchards by combining laser scans (LiDAR) and camera images to find and measure fruits on trees (Figure 2). It is set up to follow the same fruits across multiple visits, tracking their growth over time. In field tests, it correctly counted nearly all the 1,790 apples across 60 trees, estimated fruit size within just over a centimeter, and was significantly more accurate at tracking fruit growth over time than existing methods. In essence, the system provides an active “perception system,” enabling autonomous exploration of unknown environments.
Explanation/Background
NSF IoT4Ag researchers validated the technology through real-world experiments as well as simulations with multiple robots, developing multiple tools and approaches that advance overall capabilities for precision farming. Among them is an open-source autonomous UAV hardware–software system designed for long-range navigation that addresses the unique challenges of perception and control beneath dense foliage. Another is TreeScope, a robotics dataset for agricultural environments that provides important data and detailed models of plants and trees, which can be used to estimate crop yields and check plant health. Building on Treescope’s foundation, the researchers also developed SideSLAM, a decentralized real-time system for metric–semantic mapping by heterogeneous multi-robot teams. They then extended that work to construct a fine-grained mapping framework capable of estimating tree diameter profiles in under-canopy conditions.
Location
Philadelphia, Pennsylvaniawebsite
Start Year
Microelectronics and IT
Quantum, Microelectronics, Sensing, and IT
Lead Institution
Core Partners
Fact Sheet
Outcome/Accomplishment
Understanding plant health is essential for making key farming decisions, like how much fertilizer or pesticide to use and how much produce to expect at harvest. A system of novel systems for keeping track of individual trees and even single fruits in an apple orchard during the growing season has been developed by researchers at the National Science Foundation (NSF)-funded Internet of Things for Precision Agriculture (NSF IoT4Ag) Engineering Research Center (ERC), headquartered at the University of Pennsylvania.
Location
Philadelphia, Pennsylvaniawebsite
Start Year
Microelectronics and IT
Quantum, Microelectronics, Sensing, and IT
Lead Institution
Core Partners
Fact Sheet
Impact/benefits
NSF IoT4Ag’s system uses uncrewed aerial vehicles (UAVs) to navigate under foliage canopies, which is essential for tracking changes in crops—including tree crops (Figure 1). The system creates detailed 3D and 4D maps of orchards by combining laser scans (LiDAR) and camera images to find and measure fruits on trees (Figure 2). It is set up to follow the same fruits across multiple visits, tracking their growth over time. In field tests, it correctly counted nearly all the 1,790 apples across 60 trees, estimated fruit size within just over a centimeter, and was significantly more accurate at tracking fruit growth over time than existing methods. In essence, the system provides an active “perception system,” enabling autonomous exploration of unknown environments.
Explanation/Background
NSF IoT4Ag researchers validated the technology through real-world experiments as well as simulations with multiple robots, developing multiple tools and approaches that advance overall capabilities for precision farming. Among them is an open-source autonomous UAV hardware–software system designed for long-range navigation that addresses the unique challenges of perception and control beneath dense foliage. Another is TreeScope, a robotics dataset for agricultural environments that provides important data and detailed models of plants and trees, which can be used to estimate crop yields and check plant health. Building on Treescope’s foundation, the researchers also developed SideSLAM, a decentralized real-time system for metric–semantic mapping by heterogeneous multi-robot teams. They then extended that work to construct a fine-grained mapping framework capable of estimating tree diameter profiles in under-canopy conditions.