Abstract: We are entering the era of devices that are able to see and understand the world. Computer vision scientists and engineers who create these future systems, from automotive safety to augmented and virtual reality, need to push the limits of embedded low power computing. We will discuss the implementation of such systems, using examples such as itSeez3D, the first mobile color 3D scanner for consumers. We will cover OpenCV as a tool for developing embedded realtime algorithms, as well as OpenVX -- the recently standardised hardware abstraction layer for computer vision.
Bio: Victor is an entrepreneur with an extensive background in computer vision. Prior to co-founding Itseez, Victor worked as a project manager and senior research scientist at Intel Corporation. He is the author of more than 25 papers in the areas of computer vision and machine learning as well as of several US and international patents. Victor has also been involved in several open source projects, being a developer of OpenCV library. Since 2012 Victor serves as chairman of the OpenVX working group at Khronos that develops the open standard for the computer vision industry.
Abstract: After a short introduction to the University of the Bundeswehr Munich (UBM) Joe Wuensche will briefly review 25 years of history of robot cars at UBM, especially achievements with the 1st and 2nd generation UBM cars, which drove autonomously, solely vision-based on German autobahns at rather high speeds already back in 1986. Then he will focus on his current research vehicle MuCAR-3, a modified VW Touareg used for autonomous on- and off-road driving. Joe will explain the fundamentals of the 4D-approach to environment perception developed at UBM, followed by a closer look at visual and Lidar-based perception inclusive various sensor-fusion results. Situation assessment and off-road object-related navigation without GPS or with only very sparse GPS information is another focus of the talk.
Throughout the talk, videos will demonstrate achieved performance, especially on-board race videos from the European Land Robot Trials (ElROBs) and of various projects of automated UGVs including visual night-time perception and a new approach to recursive stereo vision. He will also highlight the differences between the ElROB trials and the DARPA Grand and Urban Challenges, where his team participated as part of the German team "AnnieWay", making it into the 2007 finals of the Urban Challenge.
Bio: After receiving his MSc in Aerospace Engineering from UT Austin, USA, Joe Wuensche was supervised for his Ph.D. by Prof. E.D. Dickmanns, a well known pioneer for autonomous cars, at the University of the Bundeswehr Munich (UBM), with graduation in 1987. He served in leading positions at MBB Automation Technology, Schroff GmbH, USA, and Pentair Europe. He has been a member and chairman of the board of several publicly trading German companies.
Since 2004 he has the chair for "Autonomous Systems Technology" at UBM, since 2008 he is the director of the identically named institute, and since 2011 he is the director and CEO of the research center "MOVE" (modern vehicles) at UBM. In addition he has been since 2009 the CEO of his own, an UBM spin-out, company IfTAS GmbH. Currently he is the elected dean of the department of Aerospace Engineering at UBM. His team operates the autonomous vehicles MuCAR-3 and MuCAR-4, with which it has successfully entered many competitions and events: MuCAR-3 came in first in every event entered during all ElROB trials from 2007 to 2012 (apart from one second place at ElROB 2008); with team AnnieWay his team was one of only 11 finalists in the 2007 DARPA Urban Challenge.
Joe's research interests focus on perception and biologically inspired navigation in challenging, unstructured off-road environments. He and his team have automated numerous unmanned ground vehicles ranging in size from small 50 kg robots, various passenger and electric logistic vehicles, to 3000 kg reconnaissance vehicles and 25 ton trucks. His research funding comes from the German Research Foundation, the German ministry of defence, and all major German car companies. Together with his approximately 15 Ph.D. students he has authored more than 100 publications.
- Richard Green
- University of Canterbury
Abstract: In recent times, automating agriculture has become increasingly dependent on computer vision algorithms. This presentation will discuss the challenges of automating precision agriculture using autonomous aerial vehicles, autonomous underwater vehicles and farm robots. As an example of such challenges, this presentation will describe a robot system for automatically pruning grape vines, which uses a trinocular system to build 3D models of vines and an AI system to determine which canes should be pruned. Within this mobile platform straddling rows, a six degree-of-freedom robot arm makes the required cuts. The main innovation is the computer vision system which builds 3D models of the vines. These models must be sufficiently complete and correct to make decisions on where to prune. This system is based on a feature matching and incremental bundle-adjustment pipeline where each stage of this pipeline is customised to work well for vines. The AI algorithm uses these models to decide which canes to cut, where these decisions are consistent with the decisions made by pruning experts. - This keynote is co-authored and co-presented by Dr. Tom Botterill, University of Canterbury.
Bio: Associate Professor Richard Green completed his PhD in 2003 from the University of Sydney focusing on computer vision based tracking and recognition of sports skills. While pursuing his PhD, he won an IEEE Journal Best Paper of the Year Award. He had previously successfully led the research, development and commercialisation of banking software for the international banking industry from his own start up company of 50 staff in Australia and eventually sold the IP to a multi-national computer corporation. Since 2004 he has been lecturing in computer science at the University of Canterbury with over 150 refereed publications. He heads the Computer Vision Research Lab with an emphasis on 3D robot vision for autonomous aerial vehicles, autonomous underwater vehicles and also leads a $3 million MBIE research project which has developed an intelligent vision-based pruning system.