The evolution of autonomous robotic platforms have continued to keep the excitement around the potential and implementation of swarm farming for large scale agricultural production. A robotic platform for a specific crop production operation needs thoughtful considerations, especially when the expectation is to continually enhance agricultural productivity, profitability, and possibly, environmental sustainability. To realize swarm farming goals, these autonomous robotic platforms will require seamless integration and the functioning of numerous sub-systems.
The navigation and guidance of an autonomous platform, whether it is designed to operate within a row for in-row operation or for multi-row, needs to be robust to operate at optimal speeds without causing any crop damage. Platform design and navigation technology is particularly critical considering the large field sizes, terrain, crop residue, varying canopy structures, and lighting conditions. Additional sensors and models need to be embedded for advanced diagnostics/prognostics for preemptive maintenance, automatic re-calibration, and autonomous decision making to return to base station instead of getting stuck in the middle of the field. Knowledge is rapidly increasing around the aforementioned domains but agricultural workspace is complex, dynamic, and ever-evolving, so it requires more investment and concerted effort.
Next, the application system for these robotic platforms needs to be innovative, intelligent, and rugged. As we discussed in the previous blog, not every robotic platform system is being designed with considerations of application systems. Similar to platforms, application systems would need sensors and actuators to go beyond the capabilities of current application systems. Operators often see a knocked off spray nozzle and dislodged seed tube vigorously wasting precious crop input. In the absence of human intelligence in the cab, all relevant and practical data would be captured by an array of sensors. Computer vision, artificial intelligence, machine learning, and decision models would be required to implement site-specific application. Considerations have to be given to power utilization, processing capabilities, data storage, cost, hardware scalability, ease of integration options, serviceability, and data formats (among others).
Autonomous systems utilizing an array of sensors would generate gigabytes of real-time data at very high spatial resolution. For platforms to work in a common place in large spaces, a robust communication architecture and protocol needs to be in place. The communication would provide a continuous stream of knowledge for the platforms to co-exist and co-operate safely. In rural locations where connectivity could be a constraint, newer data communication and data handling strategies need to be developed for autonomous platforms to communicate with each other and with the master control facility. The communication system is vital for robust monitoring of platform location, operation, data sharing, and coverage/route modifications (among others). Communication systems provide knowledge of operating platforms which will help maintain safety of platforms (not collide with each other), humans, property, livestock, and allied systems working in the operational space.
Large high spatial data sets are probably considered the most valued asset in the agricultural world. Since swarm farming will be in a very intense data generation space, data security and computing would be yet another vital piece of the puzzle. If field scale communication can be established, fog/edge computing systems can do their magic and possibly upload data to the central servers either at the farm office or the cloud. To safeguard data privacy and business interests, cyber-security systems would be needed to safeguard user interest, provide customized service, and maintain business vitality.
The last caveats will be the business model for swarm farming. Questions producers would pose would include:
Is it going to be its own and operate or will it have a service contract with a company to pay for time of usage?
Will there be dedicated 24/7 service?
Who is going to own the data: companies providing service or solely belonging to the farmer?
What is the economic advantage to justify the switch?...and more!
The USDA-NIFA funded project under the National Science Foundation’s National Robotics Initiative Program at Kansas State University has several facets of above-mentioned challenges. The project team understands that all these challenges are independent yet all of them have been kept in consideration while setting goals and milestones to effectively succeed in this arena.
Bottomline, the sheer scale of agricultural fields needs mobility, intelligence, connectivity, protocols, work environment safety, and data security to operate 24/7.
By: Dr. Ajay Sharda, Associate Professor, Precision Ag/Machine Systems Engineer, Kansas State University
Dr. Ajay Sharda in the Biological and Agricultural Engineering department at Kansas State University has a USDA-NIFA funded project under the National Science Foundation’s National Robotics Initiative Program. This is one of the many projects within what he calls the FARMS (Fusing Automation and Robotics for Ag. Machine Systems) Lab at K-State. The project aims for designing a robotic platform, along with an application system, which can not only provide an alternative but add substantial intelligence in terms of artificial intelligence to conduct knowledge based real-time application.
