bio Tomas Pajdla is Assistant Professor at the Czech Technical University in Prague. He works in geometry and algebra of computer vision and robotics with emphasis on non-classical camera systems, 3D reconstruction and industrial vision. T. Pajdla published more than 75 works in journals and proceedings and received awards for his work OAGM 1998, BMVC 2002 and ICCV 2005. He served as a programme chair of ECCV 2004 and more than ten times as area chair of ICCV, CVPR, ECCV, ACCV, ICRA and BMVC. He is a member of the ECCV Board, a co-editor of IPSJ Transactions on Computer Vision and Applications and was a co-editor of special issues of the International Journal on Computer Vision and Robotics and Automation Systems. T. Pajdla has connections to planetary research community through EU projects with NASA and EADS Astrium and to automotive industry via Daimler AG.

The workshop will host an invited talk on the outcomes so far of the FP7 European collaborative Project PRoViDE - Planetary Robotics Vision Data Exploitation". This project aims at assembling imaging data from vehicles and probes on planetary surfaces to make it available as a unique database featuring fusion between orbital and 3D ground vision data and a multiresolution visualization engine.

bio Michael Milford is a Senior Lecturer and ARC DECRA Fellow of the Robotics, Vision and Sensor Networking Lab at Queensland University of Technology (QUT) in Brisbane, Australia. He received his PhD in Electrical Engineering from the University of Queensland. He was awarded an inaugural Australian Research Council Discovery Early Career Researcher Award in 2012 and Microsoft Faculty Fellowship in 2013. He has worked at the Queensland Brain Institute and his current research interests at QUT include biologically inspired robot-vision mapping and navigation, among others.

The brain circuitry involved in encoding space in rodents has been extensively tested over the past thirty years, with an ever increasing body of knowledge about the components and wiring involved in navigation tasks. The learning and recall of spatial features is known to take place in and around the hippocampus of the rodent, where there is clear evidence of cells that encode the rodent's position and heading. RatSLAM is a primarily vision-based robotic navigation system based on current models of the rodent hippocampus, which has achieved several significant outcomes in vision-based Simultaneous Localization And Mapping (SLAM), including mapping of an entire suburb using only a low cost webcam, and navigation continuously over a period of two weeks in a delivery robot experiment. RatSLAM has performed vision-based mapping on platforms including passenger cars, Pioneer and Guiabot robots, remote control cars, quad-rotors and autonomous tractors. The work has also led to recent experiments demonstrating that impressive feats of vision-based navigation can be achieved at any time of day or night, during any weather, and in any season using visual images as small as 2 pixels in size. I will discuss the insights from this research, as well as current and future areas of study with the aim of stimulating discussion and collaboration.

bio Mitch Bryson is a postdoctoral research fellow and lecturer at the Australian Centre for Field Robotics (ACFR) at the University of Sydney, Australia. He received his PhD in robotics from the University of Sydney in 2008. His past experience includes the development of vision-based navigation and mapping systems for Unmanned Aerial Vehicles (UAVs) and he currently works on Autonomous Underwater Vehicles (AUVs) within the marine robotics research group at the ACFR. He has almost ten years experience in the development and fielding of robotic systems for environmental monitoring applications including ongoing projects with Australia's Department of Primary Industries and Integrated Marine Observing System. His research interests include robotic perception, navigation, mapping and low-cost autonomous systems in support of environmental science.

Terrestrial and marine ecosystems face a variety of pressures from human impacts such as urban development, agriculture and overfishing and are expected to face increasing pressures due to predicted global climate change. Environments such as coral reefs have significant economic value worldwide; long-term monitoring of these habitats is necessary for understanding changes and the effect of habitat protection measures. Traditionally monitoring programs have relied on routine map building performed using high-flying surveys with manned-aircraft, through satellite remote sensing or ship-bourne sonar bathymetry. These technologies are often limited in their ability to complete the picture at fine-scales, owing to limits in both the spatial and temporal resolution of the data they gather. Autonomous robotic platforms such and Unmanned Aerial Vehicles (UAVs) and Autonomous Underwater Vehicles (AUVs) are technologies that have recently been used to complete the picture in these applications. These platforms provide a unique perspective on the environment, venturing to depths and close proximity to terrains that are unsafe for manned platforms, providing rich levels of detail, with the potential for high endurance and persistent mapping. In this talk I will discuss past and on-going work in the use of UAVs, AUVs and vision-based navigation and mapping technologies within several environmental monitoring projects in Australia. Owing to the smaller size of autonomous platforms and the often low-cost nature of environmental monitoring programs, vision is an ideal sensor that has been used extensively. Future research challenges in the use of these systems include real-time perception and classification in unstructured environments and the development of autonomous decision-making systems that are "aware" of the high-order scientific goals as part of a monitoring mission.

Our principal goal is the creation of a framework that allows vehicles to perform autonomous navigation in urban environments due to the awareness of suitable semantic information. To this end we are currently developing a novel approach that consists in creating semantically rich 3D maps, which encode all the information required by the navigation tasks. This map contains critical information such as: the segmentation of the voxels in different urban classes (road, building, sidewalk, fence, etc.), traffic intersections, traffic signs, quality of the terrain, and much more. Since "semantics" are hard to compute, the creation of these maps is carried out off-line and then, this information can be accessed on-board efficiently when a new vehicle needs it. Building this type of maps represents a challenge for several research lines within the computer vision field, specifically for semantic segmentation, object detection, 3D mapping, localization and overall for scene understanding. In this talk we explain our ongoing for this challenge.

• the CVVT2010, organized at ACCV2010 in Queenstown (New Zealand)
• the CVVT2011, organized at ICCV2011 in Barcelona (Spain)
• the CVVT2012, organized at ECCV2012 in Firenze (Italy)

CVVT 2011

CVVT 2012

CVVT 2012