Call for Expressions of Interest
2 PhD Positions in Robotics and Artificial Intelligence
Advanced perception and control for precision agriculture
Institute of Cognitive Sciences and Technologies, Italian National Research Council
Sapienza University
Rome, Italy
_________________________________________________
Research topic:
One of the world’s greatest challenges is to ensure sufficient food production for an ever-growing population. This is in contrast to the reduction of arable land, and the need for ecologically compatible production. Only through advanced precision farming techniques it will be possible to ensure increased production with reduced inputs of water, fertilisers and chemical plant protection products. The projects aim to study robotic and AI systems capable of minimising or completely eliminating the use of chemical agents in agriculture. Specifically, the aim is to automate precision spraying techniques, which have been shown to reduce the use of pesticides or herbicides by up to 90%. In addition, the project will assess the impact of completely replacing chemical agents with mechanical interventions using robotic actuators.
The fundamental component of a precision spraying system is the ability to accurately identify the areas to be sprayed, assessing surfaces and volumes so as to be able to minimise the amount of chemicals applied while preserving the effectiveness of the intervention. It is therefore a computer vision problem for the recognition of the parameters of interest and for 3D reconstruction, which presupposes the ability to observe from several perspectives
and to integrate the different observations for a more accurate estimate.
In order to optimise the observation process, an active monitoring approach is proposed (possibly exploiting a multi-robot system), whereby observation points are determined based on the need to acquire information relevant to both recognition and estimation of the parameters of interest. Techniques based on information theory will be studied to minimise the uncertainty related to the observed objects, and to determine future actions and observations based on the expected information gain.
These fellowships are available in the context of the projects AGR-o-RAMA, E-CROPS and with support from the Italian Ministry of Research on the FSE REACT-EU fundings.
_________________________________________________
Skills and background knowledge:
– Advanced methods for object detection and semantic segmentation
– Recurrent convolutional neutral networks, LSTM
– Active vision
– Information theory
– Bayesian inference
– Information-based motion planning
_________________________________________________
Application / Enquiry:
Expressions of interest should be submitted by October the 15th, 2021
The formal call for the studentship will be released on October 11th, 2021, however those who are interested in the position could informally contact Dr. Vito Trianni (vito.trianni@istc.cnr.it) for discussing details about the project or the admission process.
The admission process will take place at the beginning of November 2021.