Conference Agenda

Cerberus is an EU-funded research project with the objective of deploying a systematic procedure to enhance crop protection by bringing together satellite images, field-set automatic traps, proximal sensing monitoring and citizen observations to generate risk maps and enhance decision support to act earlier and with better precision, resulting in more sustainable management of crops. Focused on Mediterranean crops (vineyards, olives, and citrus) the project uses a cloud-based platform designed for fruit producers. The system automatically processes Copernicus images, combines vegetation and tree‑stress indicators with insect‑trap data, and generates pest‑pressure risk maps that support the creation of targeted spray prescription maps. Such maps are easily downloaded from the platform and uploaded to an intelligent blast sprayer developed along the project, which is capable of delivering tree-adapted variable rates while maintaining a constant pressure of 6 bar, ensuring a stable droplet size. The sprayer features six sectors independently controlled by a computer and 32 PWM solenoid-actuated nozzles. Preliminary results conducted on the pilot plots of the project in Spain, Italy and Cyprus reveal a satisfactory performance of the sprayer, a similar biological efficacy comparable to conventional systems, and important reductions in pesticide use.

Reliable sensors for measuring canopy characteristics and technologies enabling real-time Variable Rate Application (VRA) spraying are currently available. Nevertheless, a key challenge remains the development of algorithms able to translate sensors outputs into appropriate spray volumes rates without compromising crop protection efficacy. This study aimed to determine, through field experiments, the optimal spray volume rate for different vine canopy densities. Trials were conducted at three vegetative stages (BBCH 55, 57, and 77), testing four spray volumes rates per stage. Spray quality was evaluated by measuring coverage and impact density using water sensitive papers placed within the canopy. Results were compared with pre-established efficacy thresholds to identify the most suitable application rates according to canopy characteristics. The identified volumes increased proportionally with canopy density throughout the season and ranged from 130-450 L ha^-¹ and 90-350 L ha^-¹ for fungicides and insecticides, respectively. These outcomes were integrated into the control algorithm of the VRA prototype sprayer developed within the PNRR-AGRITECH project. The sprayer, equipped with Pulse Width Modulation valves, continuously adjusts the application rate according to real-time canopy volume measured by an on-board stereoscopic camera. The prototype’s performance (spray mixture savings, distribution quality, ground losses) will be presented at the conference.

Heroic viticulture is practised in highly constrained environments characterised by steep slopes, terraced vineyards, and limited mechanization, where plant protection operations are predominantly manual and associated with high physical workload and operator exposure to pesticides. Unmanned Aerial Spraying Systems (UASS) represent a promising alternative. This study reports the results of experimental trials using a DJI Agras T25 in heroic viticulture in a terraced vineyard in Massa (Tuscany, Italy). Spraying was performed to assess different operational configuration (flight mode, droplet size and application volume) in terms of canopy coverage, spray deposition, ground losses, and spray drift. A backpack sprayer was included as a reference method. Regarding canopy deposition all trials conducted with UASS have demonstrated a lower deposition in relation to backpack sprayer. In terms of canopy coverage, the trials conducted using the UASS which flew along the direction of vine row and with lower droplet size reached similar values to traditional method. Concerning the spray losses, both for spray drift and for ground losses the back sprayer guaranteed less spray losses. In conclusion, UASS technology has achieved levels comparable to conventional techniques in terms of canopy coverage. However, the issue of excessive ground losses and spray drift remains unresolved.

Sustainable agriculture necessitates transitioning the application of resources, such as irrigation and spraying, from a uniform approach to a site-specific one. Existing crop inspection tools help with this endeavour but rely heavily human intervention, making frequent data collection cost-prohibitive. Furthermore, the rising interest in agrivoltaics requires granular study to assess its impact on crop development.

To address these challenges, a fully autonomous robotic system is deployed in the agrivoltaic pear orchard of Wageningen University & Research, which comprises both a section with overhead PV panels and a control plot. The solution utilises a Clearpath Husky UGV operating on ROS2 Humble, equipped with an EdenCore Viewer, an AI-powered camera system for crop recognition.

Navigation is achieved by fusing GPS and IMU data, while a LIDAR ensures obstacle avoidance and localisation stability in GPS-disrupted environments. An energy management system, driven by a Kalman Filter and AprilTag-guided docking, enables autonomous wireless charging at Wibotic stations.

By automating the monitoring of growth metrics, the robot facilitates the precise comparative analysis of crop development between the PV-covered and control environments. This framework establishes a validated baseline for future work, including targeted irrigation and precision spraying to optimize resource efficiency.

An harvester prototype for high density olive orchards (Row-Side Continuous Canopy Shaking Harvester – RSCCSH) was developed between 2009 and 2014. The harvester comprises two symmetrical units trailed by a farm tractor. Each harvester has a vibratory rotor with flexible rods, a catching platform with conveyors belts and a temporary storage. Performance evaluation of the Row-Side Continuous Canopy Shaking Harvester (RSCCSH) prototype revealed greater quantity of fruit harvested in the areas of the canopy in contact with the rods. Given that the rods penetration in the canopy is less in the inner zone, authors were made a trial to evaluate the influence of the pruning interventions performed in the inner zone of the canopy on the prototype’s harvesting efficiency. The trial was done in a high density olive orchard of the portuguese ‘Galega vulgar’ variety submitted to the following treatments: T1 – trees with a central part of the canopy without branches (empty center) and T2 – trees with a “V” shape in the canopy central part. The orchard was pruned in spring 2020 and harvested in the following campaign (2021). No significant differences in harvesting efficiency were found between the pruning treatments.

The increasing adoption of Agriculture 4.0 technologies has led to a significant evolution of variable-rate application systems, highlighting the need for reliable tools to comparatively evaluate different technological solutions prior to their field implementation. In this context, Software-in-the-Loop (SIL) testing represents an effective approach for systematically analysing the performance of spraying systems, reducing the time, cost, and irreversibility associated with traditional experimental trials.

This work develops a SIL-based methodology for comparing four different spraying technologies, ordered by increasing logical and mechanical system complexity. The simulations were implemented in an environment based on IPG CarMaker, used as the vehicle dynamics simulation platform, and AgriSI©, employed as an agricultural simulator for field management, prescription maps, and distribution logics. The technologies were evaluated across six field scenarios characterized by varying geometric complexity and spatial variability, using performance indicators such as the distribution error with respect to the prescription map, which is a key factor for field productivity.

Subsequently, the analysis was extended to the study of non-conventional operating trajectories, aimed at reducing working time and fuel consumption. A curve distribution correction algorithm was developed to compensate for localized overdosing effects.