A new study led by Laurence Roberts-Elliott, postgraduate research student at the University of Lincoln, shows that fleets of multiple small robots will be able to identify problematic soil much quicker than traditional sampling techniques.
The proposed system would consist of a fleet of robots working in conjunction to collect and analyse soil penetration resistance data to create a map displaying where soil compaction is significant. If successfully introduced, the team says that speedier soil compaction mapping could enable areas of high soil density to be treated sooner and less wastefully.
Laurence commented, “Soil compaction has serious negative effects on the yields and nutritional value of crops and increases the emission of greenhouse gases from soil. Field-wide land management to address this can be expensive in labour, time, and fuel costs and wasteful compared to more precise and localised land management.”
The 2-D simulation presented in his study dynamically creates and allocates multiple sampling tasks to the robots while maximising the information received from each sample to significantly increase the overall efficiency of the process.
He added, “Our lightweight multi-agent simulation has allowed for rapid experimentation to quickly discover and develop effective methods of task allocation and dynamic sampling for multi-robot soil compaction mapping.”
To determine if it would be possible to introduce the system into practical agricultural systems he stressed that the technique would need to be tested in more realistic simulations with real-world experiments.