How a client leveraged our multi-camera solution to
automate farming activities
Smart farming is getting traction around the world to maximize yield and optimize resources like water. Also, there is a lot of repetitive labor-intensive skilled work involved in agriculture that gets regularly automated by machines. Cameras are important in smart agriculture both in analytics of field and also basic automation activities. Real-time analytics of the field and crop status helps in planning and management of the crop field. This case study is of a customer wanting to automate a particular part of his farming process.
Challenges of accelerating
The customer, a technology company offering automation solutions to farms was interested in a vision solution that should solve two key problems at different stages of the crop life. The customer had a mobile rover which would go through the field and needed the camera for the following purposes,
- Early Crop Stage - There was a requirement to inspect the crop regularly for presence of larvae, insects, and weeds.
- During harvest stage - Check for the stage of the fruit and pick them on time before they become overripe and also sort them based on quality.
The major challenges were the camera had to be on a moving rover and had to capture really good quality images in color for the algorithm to be able to process and make the right decisions.
Support for NVIDIA Jetson
Selection of the camera module
e-con Systems™ offered a solution with two cameras for the rover. The 2MP e-CAM20_CU0230_MOD color camera module that can capture good color images in relatively low light conditions. The images from this camera could be used by an algorithm that is dependent on the exact color and texture of the fruits. The other camera was a monochrome module See3CAM_20CUG that has a Global Shutter camera sensor and can capture images at a high frame rate without any motion blur. The images from this sensor was used by the algorithm to detect various problems in the early crop stage by fast scanning of the field.
Avoid Motion Blur
Processing Support for
How e-con Systems™ delivered a
The team at e-con Systems™ worked with the customer and after initial study and experiments decided along with the customer that 2 cameras were required for the solution. The camera modules were identified and the required proof of concept with algorithms was done for the use case as well. The positioning of the camera in the rover along with the interfacing with the NVIDIA Jetson processing unit was designed. For practical implementation purposes, the color camera module was interfaced over the MIPI interface using an IPEX cable and the See3CAM_20CUG was used as is over the USB interface for the monochrome camera. The monochrome camera was supposed to be moved around the rover and tried in different positions during the field trials until a position was finalized. The USB Interface and cabling provided this flexibility in the implementation.
The prototype design was completed and integrated with the rover and field trials were done. This solution is in mass production at present.
According to MarketsandMarkets, the agricultural robot market is projected to grow from USD 4.6 billion in 2020 to USD 20.3 billion by 2025; it is expected to grow at a Compound Annual Growth Rate (CAGR) of 34.5% from 2020 to 2025.