In this paper a solution to UAV reduced endurance and autonomous flight is proposed. With a complete on-board solution, based on artificial vision, the developed system is able to autonomously take off, navigate and land, recharging its battery by using a dedicated landing platform, both in indoor and outdoor scenarios. The landing platform includes a passive centering system to correct the landing error of the UAV, with a novel design wich reduce cost and increase the safety (thanks to small and isolated electrical contacts) without invasive hardware changes on the drone. The developed vision algorithm provides a fast and accurate measurement of UAV position with respect to the landing platform using a visual target, but at the same time it is able to automatically switch to an estimation of position that is independent from the visual target. This aspect is used during navigation or when the tracking of the target fails, ensuring a continuous position measurement feed to the controllers. The developed control system manages all the different phases of a mission (motor turning on/off, take off, navigation, landing, . . . ) with low control error, ensuring a landing over the landing platform with an error that is lower than 5cm for both x and y axis. The developed software in ROS environment is modular and provides input/output interfaces to receive command, or send data.

The capability to instantiate a cooperation among heterogeneous agents is a fundamental feature in mobile robotics. In this paper we focus on the interaction between Unmanned Ground Vehicle (UGV) and Unmanned Aerial Vehicle (UAV) to extend the endurance of UAV, thanks to a novel landing/recharging platform. The UGV acts as a docking station and hosts the UAV during the indoor/outdoor transition and vice-versa. We designed a platform and a robust landing target to automate the fast recharge of UAV. The synchronization and coordination of cooperation is managed by a Ground Control Station (GCS) developed using a versatile software toolchain based on the integration of Stateflow, auto-generation of Ccode and ROS. All the software components of UAV, UGV and GCS have been developed using ROS. The obtained results show that the UAV is able to land over the UGV with high accuracy (<5cm for both x and y axis) thanks to a visual position estimation algorithm, also in presence of wind (with gust up to 20-25km/h), recharging its batteries in a short time to extend its endurance.

The research reported in this paper proposes an Advanced Process Control system, denoted "i.Process | Steel -RHF", oriented to energy efficiency improvement in a pusher type billets reheating furnace located in an Italian steel plant. A tailored control method based on a two-layer Model Predictive Control strategy has been created that involves cooperating modules. Different types of linear models have been combined and an overall furnace global linear model has been developed and included in the controller formulation. The developed controller allows handling all furnace conditions, guaranteeing the fulfillment of the defined specifications. The reliability of the proposed approach has been tested through significant simulation scenarios. The controller has been installed on the considered real plant, replacing local standalone controllers manually conducted by plant operators. Very satisfactory field results have been achieved, both on process control and energy efficiency improvement. Optimized trade-offs between energy saving, environmental impact decreasing, product quality improvement and production maximization have been guaranteed. Consequently, Italian energy efficiency certificates have been obtained. The formulated steel industry reheating furnaces control method has been patented.

One of the main challenges for Unmanned Aerial Systems (UAVs) is to extend the endurance of small vehicles such as multi-rotors. Actually, Li-po batteries that guarantee a flight of about 20 minutes power this type of vehicles. The endurance can be extended by enabling vehicles to look for recharging station(s). In this paper, we propose a vision system able to detect and track a given pattern hosted on the target-landing platform. The pattern is also useful to estimate the UAV position while approaching the target or during the hovering close to the target. The paper focuses on an optimized adaptive thresholding technique that manages critical situations as changes in the scene's illumination / shadows. The developed system runs at 90Hz for processing a 752 x 480 grayscale image. Preliminary results on an NVidia Tegra Jetson K1 platform are also presented to distribute the computation between CPU and GPU.

No abstract