Abstract: Planning, the ability to imagine different futures and select one assessed to have high value, is one of the most vaunted of animal capacities. As such it has been a central target of artificial intelligence work from the origins of that field, in addition to being a focus of neuroscience and cognitive science. These separate and sometimes synergistic traditions are combined in our new work exploring the origin and mechanics of planning in animals. We will show how mammals evade autonomous robot “predators” in complex large arenas. We have discovered that depending on the arrangement and density of barriers to vision, animals appear to carefully manage their uncertainty about the predator’s location in order to reach their goal. Their behavior appears unlikely to be driven by cached responses that were successful in the past, but rather based on planning during brief pauses over which they peek at the hidden robot adversary that is looking for them. After peeking, they re-route to avoid the predator.
Bio: Malcolm A. MacIver is a group leader of the Center for Robotics and Biosystems at Northwestern University, with a joint appointment between Mechanical Engineering and Biomedical Engineering, and courtesy appointments in the Department of Neurobiology and the Department of Computer Science. His work focuses on extracting principles underlying animal behavior, focusing on interactions between biomechanics, sensory systems, and planning circuits. He then incorporates these principles into biorobotic systems or simulations of the animal in its environment for synergy between technological and scientific advances. For this work he received the 2009 Presidential Early Career Award for Science and Engineering from President Obama at the White House. MacIver has also developed interactive science-inspired art installations that have exhibited internationally, and consults for science fiction film and TV series makers.
When a flapping bat propels through its fluidic environment, it creates periodic air jets in the form of wake structures downstream of its flight path. The animal’s remarkable dexterity to quickly manipulate these wakes with fine-grained, fast body adjustments is key to retaining the force-moment needed for an all-time controllable flight, even near stall conditions, sharp turns, and heel-above-head maneuvers. We refer to bats’ locomotion based on dexterously manipulating the fluidic environment through dynamically versatile wing conformations as dynamic morphing wing flight.
In this talk, I will describe some of the challenges facing the design and control of dynamic morphing Micro Aerial Vehicles (MAV) and report our latest morphing flying robot design called Aerobat. Dynamic morphing is the defining characteristic of bat locomotion and is key to their agility and efficiency. Unlike a jellyfish whose body conformations are fully dominated by its passive dynamics, a bat employs its active and passive dynamics to achieve dynamic morphing within its gaitcycles with a notable degree of control over joint movements. Copying bats’ morphing wings has remained an open engineering problem due to a classical robot design challenge: having many active coordinates in MAVs is impossible because of prohibitive design restrictions such as limited payload and power budget. I will propose a framework based on integrating low-power, feedback-driven components within computational structures (mechanical structures with computational resources) to address two challenges associated with gait generation and regulation. We call this framework Morphing via Integrated Mechanical Intelligence and Control (MIMIC). Based on this framework, my team at SiliconSynapse Laboratory at Northeastern University has copied bat dynamically versatile wing conformations in untethered flight tests.
Alireza Ramezani is an assistant professor at the Department of Electrical & Computer Engineering at Northeastern University (NU). Before joining NU in 2018, he was a post-doc at Caltech’s Division of Engineering and Applied Science. He received his Ph.D. degree in Mechanical Engineering from the University of Michigan, Ann Arbor, with Jessy Grizzle. His research interests are the design of bioinspired robots with nontrivial morphologies (high degrees of freedom and dynamic interactions with the environment), analysis, and nonlinear, closed-loop feedback design of locomotion systems. His designs have been featured in high-impact journals, including two cover articles in Science Robotics Magazine and research highlights in Nature. Alireza has received NASA’s Space Technology Mission Directorate’s Program Award in designing bioinspired locomotion systems for the exploration of the Moon and Mars craters two times. He is the recipient of Caltech’s Jet Propulsion Lab (JPL) Faculty Research Program Position. Alireza’s research has been covered by over 200 news outlets, including The New York Times, The Wall Street Journal, The Associated Press, CNN, NBC, and Euronews. Currently, he is leading a $1 Million NSF project to design and control bat-inspired MAVs in the confined space of sewer networks for monitoring and inspection.
I will present a bio-inspired fish simulation platform, which we call “Foids”, to generate realistic synthetic datasets for an use in computer vision algorithm training. This is a first-of-its-kind synthetic dataset platform for fish, which generates all the 3D scenes just with a simulation. One of the major challenges in deep learning based computer vision is the preparation of the annotated dataset. It is already hard to collect a good quality video dataset with enough variations; moreover, it is a painful process to annotate a sufficiently large video dataset frame by frame. This is especially true when it comes to a fish dataset because it is difficult to set up a camera underwater and the number of fish (target objects) in the scene can range up to 30,000 in a fish cage on a fish farm. All of these fish need to be annotated with labels such as a bounding box or silhouette, which can take hours to complete manually, even for only a few minutes of video. We solve this challenge by introducing a realistic synthetic dataset generation platform that incorporates details of biology and ecology studied in the aquaculture field. Because it is a simulated scene, it is easy to generate the scene data with annotation labels from the 3D mesh geometry data and transformation matrix. To this end, we develop an automated fish counting system utilizing the part of synthetic dataset that shows comparable counting accuracy to human eyes, which reduces the time compared to the manual process, and reduces physical injuries sustained by the fish.
Bio: Masaki Nakada obtained a master degree in physics at Waseda University in Japan. Then, he finished PhD in computer science at UCLA and worked as a postdoc for another year, where he published a series of scientific papers. (https://www.masakinakada.com/) He devoted more than 10 years in the research of artificial life, specifically in the area of biomechanical human simulation with musculoskeletal models, neuromuscular controllers, and biomimetic vision. Previously, he worked for Intel as a software engineer. He received MIT Technology Review Innovator Award Under 35, Forbes Next 1000, Institute for Digital Research and Education Postdoctoral Scholar Award, Siggraph Thesis Fast Forward Honorable mention, TEEC Cup North American Entrepreneurship Competition in Silicon Valley, Japan Student Services Organization Fellowship, Rotary Ambassadorial Fellowship, Itoh Foundation Fellowship, Entrepreneurship Foundation Fellowship, Aoi Foundation Fellowship and winner of several Startup business competition & hackathons. He founded NeuralX, Inc (https://www.neuralx.ai/) in 2019 based on the IP he has developed over the decade of research. The company provides an interactive online fitness service Presence.fit (https://www.presence.fit/), where it combines the power of human instructor and motion analytics AI, which enables them to provide highly interactive online fitness experience.
Abstract: Robots have begun to transition from assembly lines, where they are physically separated from humans, to human-populated environments and human-enhancing applications, where interaction with people is inevitable. With this shift, research in human-robot interaction (HRI) has grown to allow robots to work with and around humans on complex tasks, augment and enhance people, and provide the best support to them. In this talk, I will provide an overview of the work performed in the HIRO Group and our efforts toward intuitive, human-centered technologies for the next generation of robot workers, assistants, and collaborators. More specifically, I will present our research on: a) robots that are safe to people, b) robots that are capable of operating in complex environments, and c) robots that are good teammates. In all, this research will enable capabilities that were not previously possible, and will impact work domains such as manufacturing, construction, logistics, the home, and health care.
Bio: Alessandro Roncone is Assistant Professor in the Computer Science Department at University of Colorado Boulder. He received his B.Sc. summa cum laude in Biomedical Engineering in 2008, and his M.Sc. summa cum laude in NeuroEngineering in 2011 from the University of Genoa, Italy. In 2015 he completed his Ph.D. in Robotics, Cognition and Interaction Technologies from the Italian Institute of Technology [IIT], working on the iCub humanoid in the Robotics, Brain and Cognitive Sciences department and the iCub Facility. From 2015 to 2018, he was Postdoctoral Associate at the Social Robotics Lab in Yale University, performing research in Human-Robot Collaboration for advanced manufacturing. He joined as faculty at CU Boulder in August 2018, where he is director of the Human Interaction and Robotics Group (HIRO, https://hiro-group.ronc.one/ ) and co-director of the Interdisciplinary Research Theme in Engineering Education Research and AI-augmented Learning (EER-AIL IRT, https://www.colorado.edu/irt/engineering-education-ai/ ).
Panel Speaker 1:
Panel Speaker 2:
Panel Speaker 3: