Photonic-based spectral reflectance sensor for ground-based plant detection and weed discrimination

Area of Science:

• Agricultural Engineering
• Optoelectronics
• Remote Sensing

Background:

• Selective herbicide application is crucial for efficient weed management and reduced environmental impact.
• Non-contact spectral reflectance measurements offer a promising approach for real-time plant identification.
• Existing systems often struggle with the detection of narrow-leaved plants.

Purpose of the Study:

• To describe a bench prototype photonic-based spectral reflectance sensor architecture.
• To evaluate the sensor's performance in detecting plants for selective herbicide spraying.
• To present experimental data at simulated farming vehicle speeds.

Main Methods:

• Developed a sensor using a three-wavelength laser diode module emitting polarized light.
• Employed a multi-spot beam generator to create 14 parallel laser beams over a 210mm span.
• Utilized a high-speed line scan image sensor to detect reflected light intensity.
• Implemented plant discrimination by calculating spectral response slope between specific laser wavelengths (635nm, 670nm, 785nm).

Main Results:

• Experimental data collected at simulated speeds of 7 and 22 km/h.
• The sensor architecture successfully performed non-contact spectral reflectance measurements of plants and soil.
• The system demonstrated the ability to detect narrow-leaved plants with a width as small as 12mm.
• Plant discrimination was achieved by analyzing spectral response slopes.

Conclusions:

• The developed photonic sensor architecture is suitable for selective herbicide spraying systems.
• The sensor's design, using a finely spaced laser beam array, enhances the detection of small and narrow-leaved plants.
• This technology has the potential to improve the precision and efficiency of agricultural spraying operations.