pdf presentation - invited talk
Vito TRIANNI
CNR, Roma, Italy
Abstract:
Precision agriculture represents a very promising domain for swarm robotics, as it deals with expansive fields and tasks that can be parallelised and executed with a collaborative approach. Weed monitoring and mapping is one such problem, and solutions have been proposed that exploit swarms of unmanned aerial vehicles (UAVs). In this talk, I will present the work performed towards the deployment of UAV swarms in the field. I will present the implementation of collective behaviours for weed monitoring and mapping, starting from parallel field coverage, moving to collaborative mapping of weeds, and finally discussing strategies for the self-organised deployment of UAVs into the field to implement non-uniform coverage strategies.
Vito TRIANNI Short bio
Vito Trianni is a permanent researcher at the Institute of Cognitive Sciences and Technologies of the Italian National Research Council (ISTC-CNR). He received the Ph.D. in Applied Sciences at the Université Libre de Bruxelles (Belgium) in 2006, a master in Information and Communication Technology from CEFRIEL (Italy) in 2001, and the M.Sc. in Computer Science Engineering at the Politecnico di Milano (Italy) in 2000. His research mainly involves swarm intelligence and swarm robotics studies, with particular emphasis on the design and analysis of complex self-organising systems and distributed cognitive processes.
Juan-Antonio ESCARENO
XLIM Institut de recherche - Université de Limoges
Biography - Juan Antonio Escareno
He received the Ph.D. degree in automatic control from HEUDIASYC Laboratory, University of Technology of Compiegne (UTC), France in 2008. He has held a post-doctoral fellowship position at International Joint Unit of CNRS, 3175 LAFMIA hosted by CINVESTAV, from 2008 to 2010. He was a CNRS project researcher at University of Technology of Compiegne, from 2010 to 2012. In March 2012, he was a visiting researcher at the French Nuclear Energy Commission (CEA), at Fontenay-aux-Roses, France. From July 2012 to October 2013, he was postdoctoral researcher at the Department of “Control and Micro-Mechatronics Systems” (AS2M) at FEMTO-ST (Franche-Comté Electronics Mechanics Thermal et Optical) UMR CNRS 6174, Besançon, France. From 2014 to 2018, he has held an associate professor position at Institut Polytechnique de Sciences Avancées at Ivry-sur-Seine. He si currently associate professor at the ENSIL-ENSCI (Limoges University) with research affiliation to the UMR CNRS 7252 XLIM research institute, with the mechatronics group (ReMIX).
pdf presentation - invited talk
Pascual CAMPOY
Universidad Politécnica de Madrid, Spain
Abstract
Unmanned Aerial Vehicles are increasing their application field to indoors, where their high manoeuvrability and agility can play an essential role. The main challenge for these indoor applications is the UAV accurate positioning and control regarding its environment and the objects to interact with (e.g. inspection and physical manipulation) in such a GPS denied environment. Such a big challenge requires to successfully exploit sensor fusion based on vision as a key sensor, that also plays and especial role not only in positioning, but also in scene recognition, see&avoid, as well as control and navigation itself. Several of the techniques that are now giving a big impulse in improving UAV autonomy for indoors are Visual Inertial Odometry (VIO), Visual Semantic SLAM, Deep Learning object recognition and localization, as well as direct Reinforcement Learning for planning and control, among others. This talk is aimed to bring together mentioned techniques with the common objective of contributing to use UAV as a versatile Aerial Robot in a huge amount of GPS denied robotics applications.
Biography - Pascual CAMPOY
Pascual Campoy is Full Professor on Automatics at the Universidad Polit ́ecnica Madrid UPM (Spain) and visiting professor in TUDelft (The Netherlands), he has also been visiting professor at Tong Ji University (Shanghai-China) and Q.U.T. (Australia). He currently lectures on Control, Machine Learning and Computer Vision He is leading the Research Group on “Computer Vision and Aerial Robotics” at U.P.M. within the Centre of Automatics and Robotics (C.A.R.), whose activities are aimed at increasing the autonomy of the Unmanned Aerial Vehicles (UAV) by exploiting the powerful sensor of Vision, using cutting-edge technologies in Image Processing, Control and Artificial Intelligence.
He has been head director of over 40 R&D projects, including R&D European projects, national R&D projects and over 25 technological transfer projects directly contracted with the industry. He is author of over 200 international scientific publications and nine patents, three of them registered internationally. He is awarded several international prices in UAV competitions: IMAV12, IMAV13, IARC14, IMAV16 and IMAV17.
Thibaut TEZENAS DU MONTCEL
PhD student at Gipsa-lab, Grenoble
Abstract
Multiple obstacle avoidance algorithm have been proposed over the past years but they have been tested according to different protocols. Some were tested statistically by repeating a task in specific simulated worlds, some were tested to avoid a number of real and strategically positioned obstacles, and another group went through outdoor flights testing in environments that are tough to characterize. The BOARR benchmark aims to give a common framework to test and compare obstacle avoidance algorithms for quadrotors. It offers multiple sensors and multiple indicators relevant to all quadrotors obstacle avoidance algorithms. It uses Ros, Gazebo and RotorS and can be easily deployed.