Semantic Knowledge for Autonomous Smart Farming

Abstract

A higher degree of automation - and autonomization - of agricultural processes is expected to lead to productivity gains, especially in light of more environmentally-friendly farming practices, while improving the safety of agricultural processes. To exploit the potential of this development, it should be possible to flexibly integrate devices and services within service mashups, and thereby enable them to provide higher-value services together, However, current farm automation tools instead tend to reinforce vertical functional silos and tight coupling within often proprietary systems that manage the farm environment information. We propose to describe capabilities of individual devices and services and interlink them across components and with the description of the farm environment. We posit that this will better enable autonomous agents - software agents as well as humans - to perform complex agricultural tasks while integrating heterogeneous devices and services across multiple vendors. Concretely, we describe - and demonstrate in a laboratory setting - the usage of a Knowledge Graph to describe the environment and equipment used to perform farming tasks. We show how a multi-agent-based automation system for smart farming uses this graph to reason about the state of the environment and the agents to plan the achievement of user-specified goals. Furthermore, we show how such knowledge-driven autonomous systems may include human agents alongside artificial agents as first-class citizens, towards realizing “Social Machines” in the agriculture domain.