Abstract

The oceans play a key role on our planet and underwater exploration is fundamental to improve our knowledge of the ocean. Understanding and managing ocean resources implies developing techniques and technology for ocean exploration beyond the capacity of human divers.

In this talk we will present a novel paradigm for ocean exploration through the development of 3D imaging techniques to be deployed from an underwater robot. However, underwater optical imaging faces more challenges than that in air, since light propagating underwater suffers from strong absorption, scattering and color shift. All these effects result in dim and blurry images, which means that optical images need first to be pre-processed. Therefore, the first part of the talk will deal with the peculiarities of using vision in a participating medium.

Next, we will describe ongoing work at the University of Girona towards the development of monocular-based 3D reconstruction methodologies. Then, we will discuss the use of these techniques to carry out surveys with a semi-autonomous robot, but now equipped with an omnidirectional underwater camera. Endowing the robot with the ability to build 3D maps of the environment in real time makes the robot more aware of its environment. This omnidirectional spatial-awareness gives the robot the capacity of ensuring its safety even while being teleoperated by an unskilled pilot (e.g., by a scientist that is exploring the ocean floor), automatically avoiding possible obstacles/hazards. At the same time, the omnidirectional camera allows the pilot to experience the virtual environment through head movements and feel immersed in a simulated environment.

Biography

Dr. Rafael Garcia received the engineering degree in computer science in 1994 from the Autonoma University of Barcelona (UAB), and the Ph.D. degree in computer engineering in 2001 from the University of Girona (UdG), Spain. His research activity mainly focuses on robotics in topics such as robot navigation and mapping, sensor fusion, 3D reconstruction, semantic representation of video imagery and large-scale optical mapping. Currently, he is the director of the Underwater Vision Lab of the University of Girona.

Dr. Garcia has been visiting researcher at the Universität der Bundeswehr (Germany), University College Cork (Ireland), IRISA-INRIA (France), and the University of Miami (USA). He is involved in several national and transnational projects in the field of robotics and computer vision. Dr. Garcia is author of 3 books, 40 journal articles, and more than 100 papers in peer-review conferences.

Abstract

While robots dominates repetitive works in factories, the design and controller of these robots are not suitable for relatively complex, unstructured tasks that humans do easily. These tasks typically require force sensing and dynamic interaction force control. Conventional robots are not built to control force or being flexible to perform like human arms. The talk will discuss how the new design paradigm allows such dynamic interactive force control with environments in the context of legged locomotion. As an embodiment of such robot design paradigm, the latest version of cheetah robot and force feedback teleoperation arms will be presented. This new class of robots will play a crucial role in future robot applications such as elderly care, home service, delivery, and services in unfavorable environments for human.

Biography

Prof. Sangbae Kim, is the director of the Biomimetic Robotics Laboratory and an Associate Professor of Mechanical Engineering at MIT. His research focuses on the bio-inspired robot design by extracting principles from animals. Kim's achievements on bio-inspired robot development include the world's first directional adhesive inspired from gecko lizards, and a climbing robot, Stickybot, that utilizes the directional adhesives to climb smooth surfaces featured in TIME's best inventions in 2006. Recent achievement includes the development of the MIT Cheetah capable of stable outdoor running up to 13mph at an efficiency of animals and autonomous jumping over an obstacles. This achievement was covered by more than 300 media articles. He is a recipient of King-Sun Fu Memorial Best Transactions on Robotics Paper Award (2008), DARPA YFA(2013), NSF CAREER (2014) award, IEEE/ASME Transactions in Mechatronics (2016) and Ruth and Joel Spira Award for Distinguished Teaching.

Abstract

Urban environments present many research challenges for robotic perception and navigation. When navigating through a city, a robot encounters critical issue arising from perceptual aliasing and a highly dynamic environment.

In this talk, we address existing issues in perception and navigation and present SLAM-based techniques for solving them. Among many solutions, we present map-based navigation using SLAM. Toward long-term and large-scale autonomy, environment invariant localization is essential. However, localization becomes exponentially difficult as the revisit occurs in a longer term and in a large area. We focus on two widely adopted perceptual sensors (i.e., cameras and LiDARs) and their robust localization with respect to the map built by SLAM. The scope includes both indoor and outdoor in timespans ranging from a day to a year. To achieve the resilience overcoming environmental variance, we tackle a multi-modal perception by exploiting sensors’ complementary characteristics. In this line of study, we introduce a direct visual SLAM, using a portable camera-LiDAR system. We will also introduce multi-modal visual localization, illustrating how this multi-modal registration can be leveraged in the urban environment. Lastly, we share and discuss some results obtained using machine learning techniques for versatile visual navigation.

Biography

Ayoung Kim is the assistant professor in the department of civil and environmental engineering with joint affiliation at KI robotics, KAIST. She received the B.S. and M.S. degrees in mechanical engineering from Seoul National University, Seoul, Korea, in 2005 and 2007, respectively, and the M.S. degree in electrical engineering and the Ph.D. degree in mechanical engineering from the University of Michigan (UM), Ann Arbor, in 2011 and 2012, respectively. She also worked as a post-doctoral researcher in naval architecture and marine engineering, UM in 2013 before works at ETRI as a senior researcher. Her research interests include visual simultaneous localization and mapping (SLAM), navigation, especially targeting long-term and large scale robotic autonomy.

Abstract

Robot-aided rehabilitation has recently been increasingly utilized to support clinicians in providing high-intensity and repetitive exercise therapy for people with neuromusculoskeletal disorders. However, its efficacy is still controversial when little is known about the neuromuscular mechanisms underlying the motor dysfunction. In this talk, I will present our recent efforts to understand mechanical impedance of human limbs and joints, which is a fundamental property of the neuromuscular system that facilitates seamless interactions with the physical environment, and to translate this knowledge into robot-aided rehabilitation. First, I will present quantitative characterization of multidimensional ankle mechanical impedance during various lower-extremity tasks and show how it can be altered due to neurological impairments such as stroke and multiple sclerosis. Next, I will introduce our recent study to characterize shoulder mechanical impedance using a novel parallel-actuated shoulder exoskeleton robot that minimally impacts natural arm dynamics. Finally, I will briefly share my perspective on how quantitative information about human mechanical impedance can facilitate the advancement of robot-aided rehabilitation.

Biography

Hyunglae Lee is Assistant Professor of the School for Engineering of Matter, Transport, and Energy (SEMTE) at Arizona State University (ASU). He currently directs the Neuromuscular Control and Human Robotics Laboratory (https://faculty.engineering.asu.edu/hlee). He received his BS and MS degrees in Mechanical Engineering from Seoul National University (SNU) and his PhD in Mechanical Engineering from Massachusetts Institute of Technology (MIT). Prior to joining ASU, he worked as a postdoctoral fellow at Sensory Motor Performance Program, Shirley Ryan AbilityLab (formerly Rehabilitation Institute of Chicago). Previously, he also worked at Korea Institute of Science and Technology (KIST) and LG Electronics. His research focuses on understanding neuromuscular factors involved in dynamic posture and movement and how these factors can be used to advance physical human-robot interaction and ultimately guide rehabilitative strategies for restoring motor function in neurologically impaired patients.