Appearance Based Spatial Cognition in Robots
Prof. Dr. Işıl H. Bozma, Boğaziçi University
June 27th, 2019 Thursday - 9:30 - 10:30
Spatial cognition in robots aims to be as comprehensible as possible in relation to the environment. Therefore, it is extremely important to be autonomous. Thus, robots can act reliably on their own to create the spatial awareness necessary to perform the given tasks and interact with their surroundings more easily. As is known, spatial cognition is defined as an extremely complex feature for biological systems. From a systematic point of view, the process of providing spatial information, storing it in an organized way, using it when necessary and updating it when necessary, requires both the operation alone and in connection with each other. In this context, apparently the information plays a huge role - as there can be many cases where metric data is not available / reliable or the computational complexity required to use it is too high. In this lecture, we will focus on advanced capabilities such as spatial identification and learning, mapping, positioning, looking and moving, environment exploration and information integration, and the approaches we have developed for the formation of these processes.
H. Işıl Bozma is Professor of Electrical and Electronics Engineering at Boğaziçi University. She received her BSc degree in Electrical and Electronics Engineering from Boğaziçi University in 1983. She received her MSc degree in Electrical Engineering from a dual degree program at Case Western Reserve University in 1985 and other MSc degree in Electrical Engineering from Yale University in 1986, and she then received the doctorate degree in Electrical Engineering from Yale University in 1992. She then established the Intelligent Systems Laboratory in 1996 and worked as the manager of this laboratory since its establishment. Between 1994 and 2000, she was a visiting researcher at the University of Michigan at Ann Arbor and from 2000 to 2007 at the University of Pennsylvania. Her awards include 2004 TESID Innovation Award, 2015 ICAR Best Original Paper Award. Her research interests include spatial cognition and mediation, multi-robots, co-operation and game theory. She has published widely in internationally active journals and conferences.
Challenges of Developing Autonomous Vehicles for Real-World Applications
Dr. Oktay Arslan, Tesla Inc.
June, 27th 2019 Thursday- 14:00-15:00
Nowadays, autonomy as a newly developing discipline has started to spread in many areas of daily life. Today's society is rapidly moving towards the use of autonomous tools that interact with people. We are witnessing significant progress in the development of these autonomous vehicles (unmanned aerial systems, space robots, driverless cars, etc.), which are designed for many different applications and have various capabilities. The security critical nature of these systems requires us to opt for algorithms and methods that are sometimes not the best in terms of performance but are simple, understandable and easy to test.
In this talk, I will share my lessons and experiences in developing various autonomy projects, from full-scale autonomous Boeing helicopters to Mars Rovers and driverless vehicles. These autonomous vehicles operate in a wide range of environmental conditions such as air, space and cities, and therefore have different capacities and capabilities. Determining and developing methods that can work in real-time on these equipment is one of the challenges of autonomy projects. I will conclude my speech with the results and videos we achieved during successful demos.
Dr. Oktay Arslan works as an autopilot planning and control engineer at the Autopilot Department at Tesla, Inc. He received his BS in Control Engineering in 2006 from ITU, Istanbul Turkey. He received his BSc degree in Computer Engineering in 2007 (double major) and his MSc degree from ITU from Defense Technology Department in 2009. Afterwards, he continued his education in the United States and received her MSc degree from Georgia Tech in Aerospace Engineering in 2012, MS in Computer Science in 2015 and PhD in Robotics Department in 2015. He worked as a research, navigation and control engineer in full-scale autonomous helicopter development project in Aurora Flight Sciences in 2014-2015, and as a researcher in autonomous site tools project in Mechatronics Group in 2015 at Aurora Flight Sciences. In the same year, he worked as a visiting researcher at MIT. After completing his Ph.D., he worked at NASA / Caltech JPL in Mobility and Robotics Systems Department in 2015-2017 for deep space exploration missions and various robotic projects. His main areas of interest are control theory, robotics, motion planning, machine learning, unmanned aerial vehicles, driverless vehicles and development of planning and control software for autonomous robotic systems.
Real-world computing: NASA / JPL Robosimian, Google Project Tango, and Entrepreneurship
Dr. Alper Aydemir, Volumental, CTO
June, 28th 2019 Friday- 9:30-10:30
In this talk, I will present three different image processing applications from research to commercialization. The first of these is NASA / JPL's Robosimian robot in the DARPA Robotics Challenge. I will begin with architectural principles that guide the various subsystems of Robosimian, including perception and mobility. I will then share my experience with a corporate research project, Google's Project Tango, Google's first 3D capable consumer phone, and building computerized vision systems. Finally, I will share my experience in these projects as an entrepreneur in the field of computer vision.
Alper Aydemir graduated from the Department of Mechatronics Engineering in 2006, from Sabanci University. In 2012, he received his PhD in robotics and image processing from the Royal Technical University in Stockholm. After getting his doctorate degree, he worked in Los Angeles, NASA / JPL robotics and image processing laboratory where he worked on driverless cars, the DARPA Robotics Challenge, and took part in a variety of other NASA projects. After NASA, he worked in the Google X lab on Project Tango, a project on 3D mapping, object recognition and scanning and virtual reality on mobile phones. Dr. Aydemir then founded a technology company, named Volumental. Dr. Aydemir still works at Volumental. Stockholm-based Volumental company, today operates in 35 countries with a team of 45 people. He was originally from Istanbul. Aydemir likes to go for long walks in Stockholm archipelago’s and talk to his friends about technology.