Confirmed Speakers

Andrew Calway, University of Bristol, UK

Robust and Efficient Visual SLAM

Visual SLAM has made huge progress over the past 10 years, revolutionising the ability of robotic platforms to map previously unseen environments and to localise themselves within those maps. Moreover, it enables high level understanding capability to be harnessed from computer vision, enabling a tight coupling of navigation and scene understanding. During this time, my lab in Bristol have made a number of contributions to the development of visual SLAM algorithms and methodologies, especially in the areas of relocalisation and efficient mapping techniques. In this talk I will provide an overview of visual SLAM and the challenges involved, followed by details of some of the key contributions that my lab have made. This includes: early work on using optimised filtering techniques (extended Kalman and particle filtering) for both point based SLAM and robust model-based 3-D tracking; scale adaptive feature matching for fast and robust relocalisation; use of higher-level structure (planes) for efficient mapping; and more recent work SLAM using RGB-D sensors, in particular the use of 3-D feature configurations and sparse representations for rapid and robust relocalisation and multiple correspondence types for improved pose estimation. I will also present some early results in the related area of place recognition in urban environments, which is a key capability for effective and scalable visual SLAM systems. The talk will aim to give both an introduction to the area for those who are new to visual SLAM, as well as giving an overview of the above research and the future challenges that face the area.

Andrew Calway's is a Reader in Computer Vision at the University of Bristol, UK, based in the Department of Computer Science. He is also a member of the Visual Information Laboratory (VIL) and the Bristol Robotics Laboratory (BRL). Prior to this he was a Lecturer at the Universities of Warwick and Cardiff, and a Royal Society visiting researcher in the Computer Vision Lab at Linkoping University in Sweden. His research covers topics in computer vision and its applications - robotics, wearable computing and augmented reality - and he works regularly with industry and on interdisciplinary projects. In recent years Dr Calway has specialised in 3-D tracking and scene reconstruction using vision, mainly in simultaneous localisation and mapping (SLAM), including relocalisation, fast and robust feature matching, and place recognition. He has published over 80 papers in international journals and conferences.

Richard Bostock, BlueBear Systems Research, UK

BlueBear Systems - Delivery and development of GPS denied solutions into operational UAS

Key to the delivery of UXV technology and systems across industry is the ability to operate with or without GPS technologies. Every delivered system requires mechanisms for both short and long term GPS denied operations, either as a result of intra-city GPS shadowing, operating indoors or intentional GPS jammers. This talk focuses on the need for GPS denied technologies, and approaches to operate in GPS denied scenarios based on BBSR Ltd experience from multi-UAS fixed wing and rotary wing UAS operations over the last 10 years. Approaches include depth based localisation (SLAM, VTL, LIDAR), GPS improvements (UTCGPS, steerable antennas), Path planning (GPS Shadowing, database), and Advanced sensor fusion (PUCE). Technologies have been demonstrated on BBSR products in RAPID, RISER and Blackstart.

Richard Bostock is a Principal Software Engineer at Bluebear Systems Research Ltd. He studied Electronic and Electrical Engineering at the University of Birmingham, and has been actively involved in the development of certifiable novel autonomous unmanned aerial systems (UAS) in complex environments for the past 8 years. Key areas of research and development include multi-UAS operator usability, mission system architectures, certification, novel sensor systems, and GPS denied operations.

Luis Enrique Sucar Succar, INAOE

Visual-based Planning under Uncertainty for 3D Reconstruction

The three-dimensional reconstruction of objets or the environment is a relevant problem for autonomous robots, including drones and mobile robots. This usually requires several views as a single view is not enough for a complete 3D model, so the robot must plan the sequence of positions of its sensor to build the model. This problem is solved iteratively, by planning the next view, which is known as the “next best view” problem. If we consider an autonomous drone or mobile robot there is uncertainty in the positioning of the sensor, which should be taken into account by the planning algorithm to obtain a robust plan. In this talk I will present a method for planning the next best view/state which takes into account the uncertainty in the positioning of the robot. It is a search based approach that generates a set of views which are evaluated according to a utility function which considers observing new surface with overlap with previous views to guarantee registration, and at the same time a path that is collision free and minimizes the distance traveled by the robot. For planning the paths we adapt the RRT algorithm, generating several paths considering the positioning error, and selecting that one that gives the maximum expected utility. The method has been evaluated in simulation and with a real manipulator mobile robot with promising results.

L. E. Sucar Succar: Luis Enrique Sucar received a PhD in Computer Science from Imperial College, London in 1992, a MSc in Electrical Engr. from Stanford University in 1982, and a BSc in Electronics Engr. from ITESM, Monterrey in 1980. He is a Senior Research Scientist at INAOE since 2006. Before he was Research Engineering at the Electrical Research Institute and Professor at ITESM, both in Mexico. He has been invited professor at Imperial College, Univ. of British Columbia, INRIA, France and CREATE-NET, ITALY. He has more than 300 publications in journals, conference proceedings and books; and has supervised more than 50 PhD and MSc thesis. Dr. Sucar is member of the National Research System Level-III, of the Mexican Academy of Science and Senior Member of the IEEE. He received the National Science Prize (Premio Nacional de Ciencias) in 2016 from the Mexican Government. His main research interests are in probabilistic graphical models, computer vision, mobile robots and biomedicine.

Rogelio Lozano, CNRS Research Director, CINVESTAV

Trajectory tracking for quadrotors and small airplanes

In the first part of the talk a quadrotor is shown to be able to localize with respect to a scene using vision. The quadrotor will perform some trajectories with respect to the scene and also stabilize with respect to the face of a person. The second part of the talk is devoted to the technique of trajectory tracking using a small airplane. Videos will be presented to illustrate the theoretical developments.

Rogelio Lozano holds a Bachelor of Communication and Electronics from IPN, Mexico. He obtained a Master in Electrical Engineering from the Research and Advanced Studies Centre (CINVESTEAV) of the IPN, a Master in Control and a PhD in Automatic Control from LAG, INPG, Grenoble. He has held several appointments: Professor at CINVESTAV; Visiting Professor at the University of Newcastle, Australia; Researcher at NASA Langley Research Center, USA; Visiting Professor at LAG INPG, Grenoble; and to date, Head of UMI 3175 LAFMIA CINVESTAV-CNRS. Prof. Lozano has participated in 2 International UAV competitions organised by ONERA and DGA in the periods 2003-2005 and 2007-2009. He has coordinated 3 UAV projects for the Picardie Region, France, in the period 1999-2011, Coordinator of a project on vertical take-off and landing planes funded by DGA in 2006-2007, and participated in a project on UAV navigation using optical flow, funded by FRAE in 2007-2011. He was also the Coordinator for UTC of an ANR project on Gun Lunched Micro Aerial Vehicles 2010-2013. His areas of interest include adaptive control and identification, dissipative dynamical systems, control of underactuated mechanical nonlinear systems, and modelling and control of mini aerial vehicles. Prof. Lozano was an Associate Editor for AUTOMATICA from 1987 to 2000, and currently, he serves as Editor of the International Journal of Adaptive Control and Signal Processing since 1993. His work on these areas has led to 22 PhD theses, 90 journal papers, 160 conferences, 4 Springer-Verlag books and 5 edited books.

Humberto Sossa Azuela, CIC-IPN

New Neural Network Models and their Training Algorithms and Applications

New Morphological Neural Networks with Dendritic Processing (MNNDP) are described in this talk. I talk about the original model and its training algorithm introduced by Ritter and colleagues in 2003, then I talk about the divide-and-conquer method described by Guevara and colleagues in 2014 and some of its applications. After that, I give a few words about our most recent modifications to the original MNNDP model and two new training methods, one based on Differential Evolution and one based on a data clustering method. I present some comparisons with several classification approaches to show some of the advantages and disadvantages of our proposals.

Dr. Sossa is an Engineer in Communications and Electronics from the Universidad de Guadalajara in 1981. He obtained a Master in Science degree with the speciality in Electrical Engineering from the Research and Advanced Studies Centre of the IPN in 1987 and a Ph.D. from the National Polytechnic Institute of Grenoble, France in 1992. Dr. Sossa is nowadays Head of Robotics and Mechatronics Laboratory of Research Centre in Computer Science of the IPN. He is a member of the National System of Researchers, level 3, member of the Mexican Academy of Sciences since 1997. He is Senior member of the IEEE, and member of ACM, INNS, AIM, among others. He is co-founder of the Mexican Academy of Computation. He has authored three books and more than 300 journal, chapter and conferences papers. He is a recipient of several prizes, among them the IPN Research Prize in 1997, 1999, 2005 and 2008, the IPN Research Diploma in 2000, the Lázaro Cárdenas Prize in the category of researcher in 2001, the Honorific Academic Award “Enrique Díaz de León” from the University of Guadalajara in 2008 and the Engineering Prize from the City Mexico in 2011. His research interest are in image analysis, pattern recognition, neural networks and modelling and control of robots.

Eduardo Morales Manzanares, INAOE

Learning to Fly with Transfer and Reinforcement Learning

Learning to perform a task can be very expensive (many training samples may be needed) and it is therefore of general interest to be able to reuse knowledge across tasks for a more efficient process. This is particularly true for aerial robotics applications due to possible damage risks. In this talk I will present two approaches of transfer learning to improve the reinforcement learning rate in aerial robotics. The proposed approaches use Gaussian processes to learn a continuous multidimensional transition function, allowing the methods to directly apply in continuous (states and actions) domains. In the first approach we transfer hyper-⁠parameter values of the source task to the target task. In the second approach we automatically generate examples for the target task based on instances previously acquired in the source task. We will experimentally show the effectiveness of the proposed approaches in reinforcement learning benchmark problems, as well as a challenging quadcopter to helicopter transfer task.

Eduardo Morales a Ph.D. from the Turing Institute -⁠ University of Strathclyde, in Scotland and an MSc in Artificial Intelligence from the University of Edinburgh. He is a member of the National Researcher System (Level 3) and member of the Mexican Academy of Science. He has been responsible of more than 25 research projects and has more than 150 peer-⁠review papers. He was an Invited Researcher at the Electric Power Research Institute, in Palo Alto, CA (1986), a Technical Consultant (1989-⁠1990) at the Turing Institute, a Researcher at the "Instituto de Investigaciones Electricas" (1986-⁠1988 and 1992-⁠1994) and at ITESM -⁠ Campus Cuernavaca (1994-⁠2005). He is currently a senior researcher at the "Instituto Nacional de Astrofísica, Óptica y Electrónica" (INAOE) in Mexico where he conducts research in Machine Learning and Robotics.

Rafael Murrieta Cid, CIMAT

Pursuit-Evasion Problems with Robots

In this talk, two pursuit-evasion problems will be presented. First, we will address the differential pursuit/evasion game of capturing an omnidirectional evader using a Differential Drive Robot (DDR) in an obstacle-free environment. The goal of the evader is to keep the pursuer farther than the capture distance for as long as possible and for the pursuer the goal is to capture the evader as soon as possible. In this talk, we will present the time-optimal strategies for each player, these strategies are in Nash equilibrium; we also analyze the decision problem of the game and present the conditions defining the winner. In the second part of the talk, another pursuit-evasion game will be described, in this second game, the problem is determining whether the pursuer, is able to maintain strong mutual visibility (a visibility notion between regions over a convex partition of the environment) of the evader in a polygonal environment. We assume an antagonistic evader who moves continuously and unpredictably, but with a constraint over its set of admissible motion policies, as the evader moves in the shortest-path roadmap, also called the Reduced Visibility Graph (RVG). We will show decidability of this problem, and provide a complexity measure.

Rafael Murrieta-Cid received the B.S degree in physics engineering from the Monterrey Institute of Technology and Higher Education, Monterrey, Mexico, in 1990 and the Ph.D. degree from the Institut National Polytechnique, Toulouse, France, in 1998. His Ph.D. research was done with the Robotics and Artificial Intelligence Group of the LAAS-CNRS. In 1998-1999, he was a Postdoctoral Researcher with the Department of Computer Science, Stanford University, USA. During 2002-2004, he was a Postdoctoral Research Associate with the Beckman Institute and the Department of Electrical and Computer Engineering, University of Illinois at Urbana-Champaign, USA. From August 2004 to January 2006, he was a Professor and Director of the Mechatronics Research Center in Tec de Monterrey CEM. Since 2006, he has been with the Centro de Investigación en Matemáticas, Guanajuato, México. In 2016 he was on a sabbatical leave at University of Illinois at Urbana-Champaign. His research interests include robotics, motion planning and control theory.

Hugo Rodriguez Cortés, CINVESTAV

Total Energy Control for Unmanned Aerial Vehicles

The concept of total energy control was introduced in the 1980s for the longitudinal control of a fixed wing aircraft. This control design strategy proposes to regulate the total energy rate and the energy distribution rate at the appropriate references to achieve the control objective. The total energy rate and the energy distribution rate are proportional to the time derivatives of the Hamiltonian and Lagrangian functions, respectively. This discussion proposes a generalization of this control design strategy for fully actuated mechanical systems. This generalization is used to design an adaptive nonlinear control strategy for the longitudinal dynamics of a fixed-wing aircraft and a control strategy to track trajectories for a multi-rotor vehicle. Experimental results are presented to illustrate the benefits of this control technique based on the concept of energy.

H. RodriguezC ortés received the B. S. degree in aeronautics from the Instituto Politécnico Nacional, the M. S. degree in automatic control from the Centro de Investigación y de Estudios Avanzados del Instituto Politécnico Nacional and the Ph. D. in automatic control and signal processing from the Université de Paris XI. He was an associate researcher at the Imperial College of Technology and Medicine in England and at Northeastern University in the USA. He is the head of the Mechatronics Section at the Centro de Investigación y de Estudios Avanzados del Instituto Politécnico Nacional. His primary research focus is design and nonlinear control of aerial vehicles.

J. Fermi Guerrero Castellanos, FCE-BUAP

Event-based collaborative control for consensus of VTOL-UAVs

Motivated by applications in physics, biology and engineering the study of consensus of collections of agents (or dynamic systems) has become an important topic in control theory. Roughly speaking, consensus means to reach an agreement regarding a certain quantity of interest that depends on the state of all agents. A consensus algorithm (or protocol) is an interaction rule that specifies the information exchange between an agent and all of its neighbors on the network. Recently, published works addressed resource aware implementations of the control law using event-based sampling, where the control law is event-driven. Such a paradigm calls for resources whenever they are indeed necessary, that is for instance when the dynamics of the controlled system varies, i.e when some events occur. In the context of cooperative distributed control (centralized or decentralized), the event-based paradigm appears as a mean to reduce the communication bandwidth in the network since, contrary to the classical scheme, an event-based control invokes a communication between the different agents only when a certain condition is satisfied. In this talk one presents the development of an event-based collaborative control applied to the problem of consensus and formation of a group of VTOL-UAVs (Vertical Take-off and Landing, Unmanned Aerial Vehicles). Each VTOL-UAV decides, based on the difference of its current state and its latest broadcast state, when it has to send a new value to its neighbors. The asymptotic convergence to average consensus or desired formation is depicted via numerical simulations and real-time experimentations.

J. Fermi Guerrero-Castellanos received a B.S. degree in electronic science, from the Autonomous University of Puebla (BUAP), México in 2002 and the M.Sc and Ph.D degree in Automatic Control from the Grenoble Institute of Technology and Joseph Fourier University, Grenoble, France, in 2004 and 2008, respectively. Between January and June 2008, he was a Postdoctoral Researcher at GIPSA-Lab Laboratory, Grenoble, France. After spending one year at the University Polytechnic of Puebla, Mexico as an assistant Professor, he joined in 2009 the Faculty of Electronics at BUAP as a full professor, where he established and directs the Control and Cyber-Physical Systems Laboratory. Currently he is also the head of Renewable Energy Engineering at BUAP. In 2016 he was a visiting research professor at the Laboratory of Image, Signal and Intelligent System (LISSI) - University of Paris-Est Créteil (UPEC). He is a Member of Mexican Academy of Science (AMC), Mexican Association on Automatic Control (AMCA) and Member of the National System of Researchers (Researcher Level I), Mexico. His research interests include guidance and control of autonomous systems, wearable robots, microelectronics systems and control of renewable energy systems.

Jose Luis Gordillo, Tecnologico de Monterrey (Campus Monterrey)

Vision oriented UAV development

Unmanned Aerial Vehicles (UAVs) is a flying vehicle capable to automatically maintaining control of its flight with no aircrew. The development and miniaturization of electronic components, in latest years, triggered the design and fabrication of UAVs, allowing at same time the miniaturization of those devices. This type of vehicles targeted the entertainment area initially, such as video grabbers or devices that could be piloted via remote control. Many different architectures were designed, among them the rotary wing such as helicopters, quadcopters, hexacopters, etc. They became popular for their capacity to take-off and land vertically and because they were able to hover, i. e., maintain its position in the air over a set point. Nowadays, UAVs are involved in many applications besides the military aspect, for example power line monitoring and inspection, terrain mapping and capture of aerial images, the latter still very popular in social events such as sports and concerts. However, this type of tasks usually requires of human intervention and depends on the skills of the pilot. This talk describes the development of UAVs architectures for providing autonomy to UAVs, based on the use of visual information. One important activity is for pose estimation using a camera out-board or with a camera in-board, using elements of environmental structures. Some other activities, like collaboration with ground vehicles and automatic landing will be described.

Jose Luis Gordillo's graduated in industrial engineering from the Technological Institute of Aguascalientes, Mexico. He obtained both the D.E.A. degree and the Ph.D. in Computer Science from the National Polytechnic Institute of Grenoble, France, in 1983 and 1988, respectively. From 1989 to 1990 he was Assistant Professor at the Department of Automatic Control of the Center for Advanced Studies and Research of the National Polytechnic Institute of Mexico (CINVESTAV-IPN). Currently he is Director and Professor at the Center for Robotics and Intelligent Systems and (CRIS) at the Tecnológico de Monterrey (ITESM). He has been a Visitor Professor in the Computer Science Robotics Laboratory at Stanford University (1993), at the Project Sharp of INRIA Rhone-Alpes in France (2002 and 2004), at LAAS-CNRS in Toulouse, France (2007-2008), and at some other universities and research institutes. His research interests are in computer vision for robotics applications, in particular autonomous vehicles, and the development of virtual laboratories for education and manufacturing. He participated and leaded R&D projects with industry like Honewell Bull in France, Sun Microsystems, Peñoles and TV Azteca; and government entities like the Mexican Army, the French-Mexican Laboratory for Computer Science (LaFMI), the Institute of the Water of the Nuevo Leon state (IANL), and the National Council for Science and Technology in Mexico (CONACyT). In particular, he promoted and leaded the National Net on Robotics and Mechatronics (RobMec). Actually he leads the Robotics National Laboratory, founded by CONACyT and ITESM, and the Robotics Research Group at the ITESM.

Bernardino Castillo-Toledo, CINVESTAV-Guadalajara

The Game of Drones

In this talk, we will present some algorithms recently applied to control of unmanned flying vehicles (drones), specially to tracking references and vision based control. We will present some applications to real problems and present and potential applications, and the trends on drones market in the forthcoming years.

Dr. Castillo-Toledo was born in Oaxaca, México in 1959. He received the B. Sc. degree in Electrical Engineering from the National Polytechnic Institute (IPN), the M. Sc. Degree from the Center of Research and Advanced Studies (CINVESTAV-IPN) and the Ph. D. degree from the University of Rome “La Sapienza”, Italy, in 1981, 1985 and 1992 respectively. He worked as a lecturer at the School of Electrical and Mechanics Engineering of the IPN from 1985 to 1989. From 1985 to 1995 he was at the Automatic Control Section of the Department of Electrical Engineering of the CINVESTAV-IPN, and since 1995, at CINVESTAV-IPN Campus Guadalajara, where he was Director from 2010 to 2015. He has held several research stages at University of Rome “La Sapienza”, University of L’Aquila and was a visiting Professor at the Laboratoire d’Automatique et d’Analyse des Systemes (LAAS) of the French Council for Scientific Research (CNR) and at University of Compiègne, among others. He has autored/co-authored more than 100 papers on journals and conferences, and has supervised about 20 PhD and 50 M. Sc. thesis. He has also led several projects on basic and applied research, some of these with industry funding. His main research interests include nonlinear control design, Hybrid systems and application of artificial neural networks and fuzzy logic techniques to control and fault diagnosis of dynamical systems and UAV’s. He is a Senior Member of the IEEE and holds the Level II of the Mexican Researchers System (SNI).

About INAOE's Accommodation and Food

Click here!

IMPORTANT DATES

Registration Deadline
February 20, 2017 (11:59 PM Mexico City Time)

REGISTRATION IS CLOSED

No more places available!