A novel method to remotely detect plant sound emission, as a response to induced stress
A novel system developed to remotely detect plant sound emission, as a response to stress. The system consists of hardware and software components: microphones that remotely pick up the plant's sounds in the ultrasonic range of 20 to 100 kilohertz; a signal processing layer; and a machine-learning algorithm that classifies the recorded sounds and reports the plant’s status. The system is based on groundbreaking research by the inventors, revealing that plants emit ultrasound signals that carry information about their state. In particular, these acoustic signals were shown to vary according to different types of stress experienced by the plant.
The system opens up a new field of precision agriculture, where farmers could listen to crops and detect various kinds of stress (due to cold, salt, drought, oxidation, pathogen attack and more) early on. In contrast to existing sensors, attached to the plant itself or placed in its immediate vicinity in order to detect changes in its condition, this innovative technology can remotely capture and interpret the plants’ sounds. This could result in much cheaper and more precise sensors than existing solutions.
Stage of Development
Profs. Hadany and Yovel, with their graduate student Itzhak Khait, found that tomato and tobacco plants made sounds when stressed by dehydration or when their stem was cut. The researchers trained a machine-learning model to discriminate between the plants’ sounds and the wind, rain and other noises of the greenhouse, and infer the plants’ status from the intensity and frequency of their acoustic emissions. The system correctly identified, in most cases, whether the stress was caused by dryness or a cut. Water-hungry tobacco appears to make louder sounds than cut tobacco, for example.
The researchers gathered evidence that other plants emit informative sounds when stressed, too. In a preliminary study, they also recorded ultrasonic sounds from a spiny pincushion cactus (Mammillaria spinosissima) and the weed Henbit dead-nettle (Lamium amplexicaule).
A patent application filed by Ramot is currently in the National Phase in the US and Europe.