Pixel Club: Visual Looming Approach for Autonomous Navigation

Speaker:
Daniel Raviv​ ​and ​Juan David Yepes (CS, Technion)
Date:
Wednesday, 5.7.2017, 14:30
Place:
Room 337 Taub Bld.

Have you ever wondered about “order amidst chaos” in a “crazy” intersection where drivers, bikers and pedestrians manage to cross it without collisions?​ This talk is about low level visual cues that can help to explain the sensing and actions of different non-colliding moving vehicles and people. The talk focuses on one of the cues, called “Visual Looming” that can be measured and act-upon locally at the moving agent level.​ ​In this presentation we will explain the initial motivation for the research, i.e., the desire to explain behaviors of school of fish and flock of birds, followed by simulated agents that self organize behavior-wise. We will share underlying principles, and how to use them to affect heading and speed. Both simulated and experimental results will be shown.​ ​The visual looming is a time-based scalar tha​​t can be obtained using raw data from one moving camera in stationary and/or changing (i.e., moving objects) environments. It does not need 3D reconstruction, and no knowledge on the direction of motion of the camera or other moving objects is needed. The value of looming is independent of the observer’s rotation.​ ​Together with another visual cue, the angular velocity, it can be used to “map” space in terms of time, using spheres and cylinders that expand and shrink, depending on specific relative motions. This cue along with others that we briefly introduce (we call them visual field invariants) remind us of cues that are used by humans to achieve basic visual tasks. In addition, logarithmic retina and fixated motion reduce the number of computations.​ ​Experiments show that it is texture independent, and can also be measured using defocusing, change in brightness, expansion of objects in the image, etc. We have used it in real driving (basic level only), and with 6DOF simulator. The looming value can also be measured by other sensors. ​​

​Bio​:
Daniel Raviv is currently a visiting professor at the Technion, researching at the Computer Science Faculty, and teaching at the Electrical Engineering Faculty.​ ​He received his B.Sc. and M.Sc. degrees from the Technion, and his Ph.D. from Case Western Reserve University in Cleveland, Ohio. He is a professor at ​​Florida Atlantic University where he is the Director of the Innovation and Entrepreneurship Lab. In the past he served as the assistant provost for innovation.​Dr.Raviv taught at Johns Hopkins University and the University of Maryland, and was a visiting researcher at the National Institute of Standards and Technology (NIST) as part of a group that developed a driverless car for the US Army (HUMVEE; 65 mph). His interest are in autonomous vehicles, active vision, teaching and learning Innovative thinking, and how to teach innovatively. Daniel developed a fundamentally different methodology for innovative problem solving a.k.a. “Eight Keys to Innovation.” He is the author of five books: Three on learning innovative thinking and two on teaching in visual and intuitive ways. He is a co-holder of a Guinness world record. ​Juan David Yepes is currently visiting the Computer Science Faculty at the Technion. He is working on Computer Vision research and simulations for vision-based navigation of multi-agent autonomous vehicles.​ ​He has a BS and MEE in Electrical Engineering and an MBA in Globlal Business. Currently he is the CEO of Industrias Terrigeno, and preparing for his Ph.D. studies. He taught specialization courses on NGN Networks. Juan has 18-year of experience in the Telecom Industry in Latin America as a senior manager in planning and operations of IP/Metro Ethernet Networks and Optical Networks. His interest interests are in Computer Vision and Autonomous Systems as well as Embedded Systems, and in developing Unity3D simulations.

Back to the index of events