BreedVision--a multi-sensor platform for non-destructive field-based phenotyping in plant breeding

Area of Science:

• Agricultural Science
• Plant Science
• Engineering

Background:

• Global food and energy security is challenged by a growing population, with crop improvement hindered by slow phenotyping methods.
• Current field-based trait measurement techniques are a significant bottleneck in advancing crop breeding programs.

Purpose of the Study:

• To develop and evaluate a tractor-pulled, multi-sensor phenotyping platform for small grain cereals.
• To overcome limitations in non-destructive phenotyping for crop improvement and genetic studies.

Main Methods:

• Integrated various optical sensors (light curtain, 3D Time-of-Flight, laser, hyperspectral, color imaging) onto a tractor-pulled platform.
• Detailed mechanical design, system architecture for data collection/processing, and phenotyping procedures.
• Conducted field trials with triticale (×Triticosecale Wittmack L.) over two years, measuring traits at three developmental stages.

Main Results:

• The platform demonstrated high technical repeatability across most sensor types under field conditions.
• Calibration for plant height determination was successful, showcasing platform application.
• The system proved highly suitable for collecting spectral and morphological plant information.

Conclusions:

• The developed multi-sensor platform provides a robust solution for non-destructive phenotyping in crop improvement.
• It lays the foundation for developing advanced sensor fusion models to quantify diverse agronomic traits.
• This technological advancement is crucial for accelerating plant breeding and achieving food security goals.