The most promising solution for achieving higher yields in agricultural products with fewer chemicals is precision agriculture. The current reactive and low-precision technologies which perform operations using an experience- rather than a data-driven “blanket approach” do not address the issue. Augmenta (one of MLSysOps partners) offers an on-tractor fully automated, low-cost system for the monitoring and full lifecycle management (e.g., for different chemical inputs such as fertilizers and pesticides) of multiple different crops.
In MLSysOps, the system will be extended to synergistically operate in real-time with a scanning system-carrying, autonomous UAV (cooperation of multiple smart edge devices). Imaging from a UAV can offer a totally different perspective from the top which can be advantageous under specific external conditions (sunrise/sunset, glares on the field, high height crops, etc.). Furthermore, AI-driven control will be used to (i) manage the software pipeline and the mapping of software modules on the heterogeneous system according to the crops, operation, and external conditions, (ii) selectively deploy the drone, according to conditions and available battery capacity, (iii) partition the computational pipeline between the on-tractor device and the drone, (iv) control network connectivity between the on-tractor device and the drone, (v) manage the storage space on the device, the type of data stored according to free capacity (raw data or detected areas of interest) and the encoding, (vi) manage data staging towards the cloud, and (vii) manage ML training tasks.