Agenda & Presentations
8.30 - 10.20:
10:20-10:50 - Coffee Break and Poster Session
10:50 - 12:30:
12:30-13:40 - Lunch Break
13:40 - 15:00:
15:00-15:40 - Coffee Break / Simulation Session
15:40 - 17:00:
Abstracts of Talks:
Mina Kamel, Roland Siegwart: Within this talk, a set of contributions in the field of robotic mapping, automated structural inspection, as well as autonomous exploration will be presented. Emphasis will be put on thorough field evaluation of the relevant systems and algorithms.
Kostas Alexis: This talk deals with the problem of autonomous exploration and mapping for facilities within degraded visual environments. A new set of methods for uncertainty-aware exploration alongside the relevant perception modules are discussed. Finally experiences from field evaluations are presented.
Sebastian Scherer: Robust localization, control, and planning are key capabilities to enable operation of flying robots in indoor environments. These environments have poor light, are dusty and require a high-level of reliability. We will present a localization, and localizability approach that is able to robustly operate in these situations. We will also present some first results on coverage planning.
Margarita Chli: This talk will discuss our latest results on the building blocks enabling autonomous navigation of a small aircraft based on visual perception and provide the outlook on the transition from a single to multi-robot collaborative estimation for a team of small aircraft.
Anibal Ollero: This presentation will have a first part dealing with new developments in aerial robotics manipulation including new platforms for dual arm manipulation and compliant manipulation, as well as topics related with control, and autonomous perception and planning. In the second part the requirements and first results on two applications will be presented: The inspection of pipes in oil and gas plants and the inspection of bridges.
Antonio Franchi: This talk will present the conception, design and development of robotic system of the MAGMaS (Multiple Aerial-Ground Manipulator Systems) type. What is probably the world’s first MAGMaS has been recently demonstrated at the 2017 Hannover Fair by an international team lead by LAAS-CNRS. Such type of heterogeneous multi-robot systems associate ground manipulators to aerial vehicles in order to benefit from the advantages of both kinds of platforms as, e.g., physical strength versus large workspace. The talk will both present the theoretical/practical challenges associated with MaGMaS and propose methods and solutions to successfully address those challenges.
Dongjun Lee: Multi-rotor drones (e.g., quadcopters, hexacopters, octacopters, etc.) are the choice of robotic platforms for many successful aerial applications, including aerial photography, mapping and geo-surveying, surveillance and reconnaissance, small packet delivery, etc. It is thus not surprising to witness many attempts to apply this drone for aerial operation and manipulation. However, we believe this drone is optimized for ease of flying, and not necessarily for other applications. In this talk, I will introduce several recent design of new robotic platforms to overcome some fundamental limitations of drones being applied for aerial operation and manipulation, including under-actuation, limited payload and dexterity, short operation time, imprecise flying, etc.
Stefano Stramigioli: In the presentation some results will be presented of physical interaction with quadcopter during flight. Example will be given of flying arms, perching and pushing on surfaces with high forces without losing flight stability.
Markus Ryll and Gianluca Antonelli: In this talk the physical interaction control for a fully actuated aerial platform is presented. To reach the fully actuation an aerial vehicle with 6 tilted propellers has been used. The interaction control manages the environment contacts, using the admittance control paradigm with the help of a momentum based wrench estimator. To generate the desired actuation wrench a specific motion control has been implemented. The validation of the proposed control has been done with experiments, in particular once tested the wrench estimator, the platform has been affected by external disturbance and a peg-in-hole task has been achieved.
Fadri Furrer, Mina Kamel: Within this talk, the overview and specifics of the RotorS Simulator will be presented. RotorS is an open source MAV gazebo simulator. It provides some multirotor models such as the AscTec Hummingbird, the AscTec Pelican, or the AscTec Firefly, but the simulator is not limited for the use with these multicopters.
Nathan Michael: Autonomous inspection with MAVs in confined and cluttered environments requires planning, perception, and control strategies that are amenable to challenging and diverse environment and flight conditions. In this talk, we will present three recent theoretic and systems developments to enable robust and precise autonomous inspection. First, we will detail a perceptual representation of the environment achieved through online learning of an approximate continuous belief distribution that represents the environment up to an arbitrary level of accuracy (contingent only on the fidelity of the perceptual sensors and the required application accuracy). Second, we will present efficient active perception strategies that enable real-time trajectory generation based on the uncertainty model associated with the approximate continuous belief distribution that yields dynamically feasible and efficient exploratory motions that reduce the environment model uncertainty while ensuring energy and time (local) optimality. Finally, we will discuss online learning and control techniques that enable trajectory tracking given substantial variations in the platform and environment conditions through an online experience-based control methodology that concurrently learns models and generates stable and convergent optimal control policies given state and input constraints as well as state uncertainty. Experimental results arising through the integration of these three techniques lead to efficient, robust, and precise autonomous inspection with MAVs.
Vijay Kumar, Giuseppe Loianno: This talk will address the estimation, control, navigation and mapping problems to achieve autonomous inspection of dark and featureless penstocks and tunnels using aerial vehicles with on-board sensing and computation. The previous tasks are challenging because range sensors do not yield adequate information and cameras do not work in the dark. A novel solution based on combination of IMU, LIDAR and cameras, robust to changes in shape of the tunnel, is proposed to solve this problem. This is the first time autonomous navigation and mapping has been achieved in a penstock without any external infrastructure. Finally, an overview of the future research directions in inspection and search and rescue with small vehicles will be provided.
Matthias Faesslner, Davide Scaramuzza: This talk addresses the topic of vision-based flight and dense mapping using micro aerial vehicles.
Matko Orsag: This talk overviews the methodology and experiments towards automated inspection of wind turbines using aerial robots.
Christos Papachristos: This talk describes a set of under development methods towards autonomous multi-modal characterization of nuclear sites in order to robotically support the waste decommissioning effort.