|Member Login

Event

2–7 February 2020

Robotic Vision Summer School (RVSS) 2020

Overview


 

The Australian Centre for Robotic Vision presents the Robotic Vision Summer School (RVSS) in Canberra, Australia. The ability to see is the remaining technological roadblock to the ubiquitous deployment of robots into society. RVSS provides a premium venue for graduate students and industry researchers to learn about fundamental and advanced topics in computer vision and robotics. The summer school also provides a unique opportunity to experiment with computer vision algorithms on actual robotic hardware.

The summer school will be held at the Australian National University (ANU) campus in Canberra, from Sunday 2 February 2020 to Friday 7 February 2020.

The program incorporates both talks and practical sessions run by world leading researchers in robotics and computer vision. This is an international summer school which offers a tremendous opportunity to broaden educational experience and professional networks for attendees.

The program is targeted at the level of later-year Masters and new PhD students in the Australian system. Attendees are expected at minimum to have a background in computer/electrical engineering of a level of an undergraduate degree or equivalent experience. Mature students with non-standard backgrounds are welcome. Attendees will be expected to undertake team based computer programming tasks in the workshop activities. Later year higher degree students will still benefit from the in-depth technical talks and face-to-face meeting with world-recognised peers and are welcome to apply, however priority will be given to early year students if numbers are tight.

If you have any questions, please email: rvss@roboticvision.org

Venue and Transport


RVSS 2020 will be held at the Australian National University (ANU) in Canberra.

Transport to Venue

Coach transportation between Sydney and Canberra will be provided for Summer School participants. Transport for travellers flying in to Canberra will be confirmed shortly.

Registering at Venue

Delegates will be provided with conference name tags as proof of registration. Name tags should be worn to all events.

Accommodation

John XXIII College, Australian National University

Meals

All meals are included in the registration cost. If you have any special dietary or religious requirements, please advise us of these requirements through your registration form.

Clothing

The weather in Canberra can be variable.  Please ensure that you bring adequate clothing for both hot and cool weather, a hat and sunscreen.  You may wish to attend a range of social activities, so please bring appropriate comfortable clothing.

For More Information Please Read

RVSS Information booklet 2020

 

Registration


*Registrations have now closed*

The school will be open to 80 qualified, motivated and pre-selected candidates. Master students, PhD students, post-docs, young researchers (both academic and industrial), and academic/industrial professionals are encouraged to apply.

Applications to attend RVSS 2020 must be received before Wednesday 18 December to receive the early-bird discount. Apply well before this date to guarantee your place and acquire your visa (allow 8-12 weeks). Applications received after 18 December will not be processed until 6 January due to the holiday break.

Pricing

RVSS 2020 Workshop TicketInclusions$AUD (Australian Dollars)
 Attendees [early-bird*]workshops, accommodation, meals and transport from Canberra$890
 Attendees [late†]workshops, accommodation, meals and transport from Canberra$950
 Attendees – No Accommodation [early-bird*]workshops, meals and transport from Canberra$715
 Attendees – No Accommodation [late†]workshops, meals and transport from Canberra$775
 Centre Attendees (includes partner institutions, chief investigators, research fellows, and Centre PhD students)workshops, accommodation, meals and transport from CanberraFees covered by the Centre
 Invited Speaker/Organiser (includes invited speakers, workshop organisers and conference committee)workshops, accommodation, meals and transport from CanberraFees covered by the Centre

* For applications received before (or on) 18 December 2019
† For applications received after 18 December 2019

Centre staff and students

All Centre-affiliated students are expected to attend RVSS in the first year of their PhD candidature. Later year Centre students should seek their supervisors’ approval before registering. Centre-affiliated staff who are presenting must still register. Other Centre staff should seek approval from their node leader before registering.

Non-attendance and refunds

If you are unable to attend RVSS for compassionate or legal reasons you must contact the organisers at rvss@roboticvision.org. Depending on the situation and how soon before the event, partial or full refunds may be possible.

Visa requirements

Please note that in order to be granted a visa to visit Australia you must demonstrate that you meet all the legislative requirements. It is strongly recommended that you apply for your visa 8-12 weeks prior to the event date. Details of visa application requirements are available on the Australian Department of Immigration’s website. The organisers are able to provide a letter of support for visa applications, but take no responsibility for visa application processes.

Program


 

RVSS 2020 program

Peter CorkeWesley ChanTom DrummondRavi GargPeter CorkeRavi GargRavi GargTom DrummondRobert Mahony



Workshop


 

Workshop Website Link

Rationale

Robotic Vision is an integrative science, bringing together different technologies, algorithms, and science for vision systems and visual analysis and making them work on a real robot in real time. The Robotic Vision Summer School provides students with an experience of the whole robotic vision system development life-cycle through a structured workshop activity involving programming a small differential drive robot to undertaking a simple but challenging visual identification task.

Teams

The workshop will be done in teams of three (if we have odd numbers, there may be one team of 2). To improve the networking aspect of the workshop we will have a team forming process on the Sunday night to try and split everyone up so you are not working with your colleagues. However, the goal of the workshop is collaborative as well as competitive and you should feel comfortable sharing ideas and even bits of code between teams during the week.

Robot


Figure 1: The PiBot Robot.

The PiBot is a differential drive two wheeled robot with equipped with a camera distributed by a QUT spinoff and designed as a teaching aid. The robot has

  • A Raspberry Pi 3B computer with a color camera
  • an i/o board with embedded processor that interfaces with motors and provides a simple UI (20×4 OLED display, pushbuttons and other LEDs)
  • ability to run code onboard written in Python or C++

The programming will be undertaken in Python in an Anaconda environment on one of the machines for the team. One of your team will need to install the anaconda environment on their laptop and then follow the provided instructions to install the Python environment for the workshop. Alternatively, we have a number of laptop computers available to loan for the week if you do not have anaconda or don’t wish to install this on your own machine.

Task

The task is to find the animals in a terrain and report their position and orientation accurately to the team – see Figure 2 below. The task is collaborative. There will be three teams robots in the terrain for each trial. At the end of the trial the three teams will have 15 minutes to collate and fuse their data to provide the best estimates of the animals positions. Each team in a trial will get the combined score for the group of three teams. There will be multiple trials with different teams. Thus, to win you need your co-teams to do well, but somehow save a little bit of extra advantage for yourself to stand out on average.


Figure 2: Two PiBots in a simulated terrain. The wooden beacons have ARUCO markers one each surface. There is a provided vision module that identifies ARUCO markers, identifies them by number and provides a bearing and distance. The animals have no identifying features and must be identified using vision algorithms. A base level CNN will be provided that can identify and classify animals and provide a rough bearing.

The terrain is an area 4m by 3m large in the workshop room. One side will be the room wall and one side will be plain pin boards, simplifying the visual background in images and improving visual identification for images taken in these directions. The other two sides will be open and will have of peoples feet, bags, tables, chair legs, movement, etc, providing complex visual background for images. The terrain will have a number (around 15) landmark beacons in unknown positions. These beacons are 10cm by 10cm blocks of wood with ARUCO markers. Code is provided that identifies each ARUCO marker uniquely and provides a bearing and distance for each ARUCO marker visible in a given image. This data is used in a Simultaneous Localisation and Mapping (SLAM) algorithm to build a map and localise the robot. A base level SLAM algorithm will be provided for this task as one of the examples in the workshop tutorials, however, there will be a lot of latitude for improvement and tuning of the algorithm during the workshop sessions.

The robot will be navigated through the terrain by remote control by a pilot who will be sitting behind a screen and will not be able to see the terrain. The pilot will operate the user interface on the team laptop for the robot and will only be able to see graphics that the team has developed during the workshop. The minimum graphics needed are provided as starting code, a visual feed from the camera and a set of navigation and image controls. Teams may choose to add additional graphics such as real time visualisation of map and localisation uncertainty, estimates of confidence in animal identification, classification, and localisation as well as additional semi-autonomous navigation capability as can be imagined.

The second team member is termed a wrangler. This team members is not allowed to view the laptop output, but will stand around the terrain and can view the robot live during the trial. They may shout instructions to the pilot, where to move, where to take images, etc. They will also be allowed certain liberties to touch the robots should they become stuck or leave the terrain.

The third team members can be allocated either as a second pilot or second wrangler as determined by the team.

The position of the landmarks, the starting position of the robots, the positions of the animals, will all be determined by the invigilators for each trial.

Identifying Animals

The core task is to identify and locate the animals in the scene. This task must be done semi-autonomously.

The identification and classification of an animal in an image must be done solely from visual data. The pilot can initiate the capture of a classification image at any time (this could be automated if the team thinks it is a good idea). An automated system must be used to identify and classify animals in the scene and their relative direction from the robot. A deep neural network will be provided for this task as one of the examples in the workshop tutorials. This basic network can be retrained during the week to provide better classification performance or better pose estimation of the animals. The data of classification and pose estimation will be saved in a standardised data format that will be provided.

It is not legal to hand label animal, or separately enter, information relating to animal identification or location from images. The actual initiation of taking an image is pilot controlled. Thus, pointing the robot directly at an animal before taking an image is legal labelling that image as one for animal detection (there will be many more images used for SLAM) is legal. However, once the image is taken, all identification, classification and pose estimation of the animal must be automated.

Evaluation

The teams are evaluated on the accuracy of their estimates of the animals.

For each image taken with an animal identified there is a standard data format provided that will save the location (and uncertainty) of the robot relative to an inertial reference frame, and the relative position of the animal. This data will be saved into a file which is then available following the trial and the three teams will combine data into a single fused data set that can be used for evaluation of the final positions of the animals. Each team will provide a map estimate with uncertainty of the beacons and these maps will be fused to provide a single map. Each identification of an animal will be an independent statistical measurement of the animal position with uncertainty in the location of the robot and the relative pose of the animal. These estimates can be fused to provide a single best estimate of the animal position. The mathematics of the data fusion problem will be presented during workshops and detailed notes will be provided. The code to fuse maps, and then fuse animal pose estimates must be developed individually by each group.

The final score awarded will be based on a least squares cost across the three animal classes that must be identified.

Speakers


Stefan Leutenegger

Stefan Leutenegger

Imperial College London

Stefan Leutenegger is a Senior Lecturer (USA equivalent Associate Professor) in Robotics at Imperial College London. He leads the Smart Robotics Lab at Imperial and co-leads the Dyson Robotics Lab with Prof. Andrew Davison. His research is centred around autonomous robot navigation: robots need dedicated sensing capabilities as well as algorithms for localisation inside a potentially unknown environment. This includes localisation and mapping with a suite of sensors, most importantly cameras, to be processed efficiently to yield accurate results at real-time.

Tobi Delbruck

Tobi Delbruck

ETH Zurich and The Institute of Neuroinformatics, Zurich

Tobi Delbruck received a Ph.D. degree from Caltech in 1993 in the inaugural class of the Computation and Neural Systems program founded by John Hopfield, as a student of Christof Koch, David van Essen and Carver Mead. Currently he is a Professor of Physics and Electrical Engineering at ETH Zurich in the Institute of Neuroinformatics, University of Zurich and ETH Zurich, Switzerland, where he has been since 1998. He co-organized the Telluride Neuromorphic Cognition Engineering workshop and the live demonstration sessions at ISCAS and NIPS. Delbruck is past Chair of the IEEE CAS Sensory Systems Technical Committee. He worked on electronic imaging at Arithmos, Synaptics, National Semiconductor, and Foveon and has founded 3 spin-off companies, including www.inilabs.com , which supports basic R&D on neuromorphic sensory processing. He has been awarded 9 IEEE awards and was named a Fellow of the IEEE Circuits and Systems Society for his work on neuromorphic sensors and processing. Besides playing with neuromorphic sensors, he likes to read, play tennis and basketball, and practice card magic on unwary subjects.

Dana Kulić

Dana Kulić

Monash University

Dana Kulić received the combined B. A. Sc. and M. Eng. degree in electro-mechanical engineering, and the Ph. D. degree in mechanical engineering from the University of British Columbia, Canada, in 1998 and 2005, respectively. From 2006 to 2009, Dr. Kulić was a JSPS Post-doctoral Fellow and a Project Assistant Professor at the Nakamura-Yamane Laboratory at the University of Tokyo, Japan. In 2009, Dr. Kulić established the Adaptive System Laboratory at the University of Waterloo, Canada, conducting research in human robot interaction, human motion analysis for rehabilitation and humanoid robotics.  Since 2019, Dr. Kulić. is a professor at Monash University, Australia. Her research interests include robot learning, humanoid robots, human-robot interaction and mechatronics.

Donald Dansereau

Donald Dansereau

University of Sydney

Dr Donald Dansereau is a senior lecturer in the school of Aerospace, Mechanical and Mechatronic Engineering, and the Perception Theme Lead for the Sydney Institute for Robotics and Intelligent Systems. His work explores how new imaging devices can help robots see and do, encompassing the design, fabrication, and deployment of new imaging technologies. In 2004 he completed an MSc at the University Calgary, receiving the Governor General’s Gold Medal for his pioneering work in light field processing. In 2014 he completed a PhD on underwater robotic vision at the Australian Centre for Field Robotics, followed by postdoctoral appointments at QUT and Stanford University. Donald’s industry experience includes physics engines for video games, computer vision for microchip packaging, and chip design for automated electronics testing.

Peter Corke

Peter Corke

Queensland University of Technology

Peter Corke is a robotics researcher and educator. He is the distinguished professor of robotic vision at Queensland University of Technology, director of the ARC Centre of Excellence for Robotic Vision and Chief Scientist of Dorabot. His research is concerned with enabling robots to see, and the application of robots to mining, agriculture and environmental monitoring. He created widely used open-source software for teaching and research, wrote the best selling textbook “Robotics, Vision, and Control”, created several MOOCs and the Robot Academy, and has won national and international recognition for teaching including 2017 Australian University Teacher of the Year. He is a fellow of the IEEE, the Australian Academy of Technology and Engineering, the Australian Academy of Science; former editor-in-chief of the IEEE Robotics & Automation magazine; founding editor of the Journal of Field Robotics; founding multi-media editor and executive editorial board member of the International Journal of Robotics Research; member of the editorial advisory board of the Springer Tracts on Advanced Robotics series; recipient of the Qantas/Rolls-Royce and Australian Engineering Excellence awards; and has held visiting positions at Oxford, University of Illinois, Carnegie-Mellon University and University of Pennsylvania. He received his undergraduate and masters degrees in electrical engineering and PhD from the University of Melbourne.

Tom Drummond

Tom Drummond

Monash University

Professor Drummond is a Chief Investigator based at Monash. He studied a BA in mathematics at the University of Cambridge. In 1989 he emigrated to Australia and worked for CSIRO in Melbourne for four years before moving to Perth for his PhD in Computer Science at Curtin University. In 1998 he returned to Cambridge as a post-doctoral Research Associate and in 1991 was appointed as a University Lecturer. In 2010 he returned to Melbourne and took up a Professorship at Monash University. His research is principally in the field of real-time computer vision (ie processing of information from a video camera in a computer in real-time typically at frame rate), machine learning and robust methods. These have applications in augmented reality, robotics, assistive technologies for visually impaired users as well as medical imaging.

Ravi Garg

Ravi Garg

University of Adelaide

Ravi Garg is an Associated Research Fellow with the Australian Centre for Robotic Vision and is part of the Australian Centre for Visual Technologies at The University of Adelaide as senior research associate since April 2014. He is working with Prof. Ian Reid on his Laureate Fellowship project named “Lifelong Computer Vision Systems”. Prior to joining University of Adelaide, he finished his PhD from Queen Mary University of London under the supervision of Professor Lourdes Agapito where he worked on Dense Motion Capture of Deformable Surfaces from Monocular Video. His current research interest lies in building learnable systems with little or no supervision which can reason about scene geometry as well as semantics. He is exploring how far the visual geometry concepts can help current deep neural network frameworks in scene understanding. In particular, his research focuses on unsupervised learning for single view 3D reconstruction, visual tracking in monocular video and weakly or semi-supervised semantic reasoning in images or videos.

Robert Mahony

Robert Mahony

Australian National University

Robert Mahony is a Professor in the Research School of Engineering at the Australian National University. He received his BSc in 1989 (applied mathematics and geology) and his PhD in 1995 (systems engineering) both from the Australian National University. He is a fellow of the IEEE and was president of the Australian Robotics Association from 2008-2011. His research interests are in non-linear systems theory with applications in robotics and computer vision. He is known for his work in aerial robotics, geometric observer design, matrix subspace optimisation and image based visual servo control.

Dylan Campbell

Dylan Campbell

Australian National University

Dylan joined the Centre as a Research Fellow at the Australian National University (ANU) in August 2018. Previously, he was a PhD student at ANU and Data61/CSIRO, where he worked on geometric vision problems, and a research assistant at Data61/CSIRO. Dylan received a BE in Mechatronic Engineering from the University of New South Wales. He has broad research interests within computer vision and robotics, including geometric vision and human-centred vision. In particular, he has investigated geometric sensor data alignment problems, such as camera localisation, simultaneous localisation and mapping, and structure from motion. He is currently looking at the problems of recognising, modelling, and predicting human actions, poses and human-object interactions with a view to facilitate robot-human interaction as part of a Centre project. You can view his personal website here.

Previous International Speakers

Previous International Speakers

2019
Laura Leal-Taixé (Technical University of Munich)
Seth Hutchinson (Georgia Institute of Technology)
Jose Neira (University of Zaragoza)
Silvere Bonnabel (Mines ParisTech)
Tarek Hamel (University of Nice Sophia Antipolis)

2018
Margarita Chli (ETH Zurich)
Vincent LePetit (University of Bordeaux)
Yarin Gal (University of Oxford)
Andrea Cherubini (University of Montpellier)

2017
Davide Scaramuzza (University of Zurich)
Simon Lucey (Carnegie Mellon University)
Javier Civera (University of Zaragoza)
Sebastien Rougeaux (Seeing Machines)

2016
Frank Dellaert (Georgia Institute of Technology)
Jana Kosecka (George Mason University)
Paul Newman (University of Oxford)

2015
Raquel Urtasun (University of Toronto / Uber)
Andrew Davison (Imperial College London)
Fredrik Kahl (Chalmers University of Technology)
Lourdes De Agapito Vicente (University College London)

Organisers


Robert Mahony

Robert Mahony

Chief Investigator

Robert Mahony is a Professor in the Research School of Engineering at the Australian National University. He received his BSc in 1989 (applied mathematics and geology) and his PhD in 1995 (systems engineering) both from the Australian National University. He is a fellow of the IEEE and was president of the Australian Robotics Association from 2008-2011. His research interests are in non-linear systems theory with applications in robotics and computer vision. He is known for his work in aerial robotics, geometric observer design, matrix subspace optimisation and image based visual servo control.

Carol Taylor

Carol Taylor

ANU Node Administration Officer

Carol Taylor is the Node Administration Officer at the Australian National University (ANU). She has previously worked in Administration at the ANU including the ARC Centre of Excellence for Free Radical Chemistry and Biotechnology within the Research School of Chemistry and at the Research School of Pacific Studies and the Australian College of Mental Health Nurses. Outside of work Carol enjoys spending time with her family, reading, dancing and having a cup of tea.

Dana Kulić

Dana Kulić

Associate Investigator

Dana Kulić received the combined B. A. Sc. and M. Eng. degree in electro-mechanical engineering, and the Ph. D. degree in mechanical engineering from the University of British Columbia, Canada, in 1998 and 2005, respectively. From 2006 to 2009, Dr. Kulić was a JSPS Post-doctoral Fellow and a Project Assistant Professor at the Nakamura-Yamane Laboratory at the University of Tokyo, Japan. In 2009, Dr. Kulić established the Adaptive System Laboratory at the University of Waterloo, Canada, conducting research in human robot interaction, human motion analysis for rehabilitation and humanoid robotics.  Since 2019, Dr. Kulić. is a professor at Monash University, Australia. Her research interests include robot learning, humanoid robots, human-robot interaction and mechatronics.

Dylan Campbell

Dylan Campbell

Research Fellow

Dylan joined the Centre as a Research Fellow at the Australian National University (ANU) in August 2018. Previously, he was a PhD student at ANU and Data61/CSIRO, where he worked on geometric vision problems, and a research assistant at Data61/CSIRO. Dylan received a BE in Mechatronic Engineering from the University of New South Wales. He has broad research interests within computer vision and robotics, including geometric vision and human-centred vision. In particular, he has investigated geometric sensor data alignment problems, such as camera localisation, simultaneous localisation and mapping, and structure from motion. He is currently looking at the problems of recognising, modelling, and predicting human actions, poses and human-object interactions with a view to facilitate robot-human interaction as part of a Centre project. You can view his personal website here.

Shelley Thomas

Shelley Thomas

Communications Specialist

Shelley joined the Australian Centre for Robotic Vision as Communications Specialist in July 2018. Convinced that everyone has a story to tell, she’s our resident ‘Chatbot’ of sorts. Shelley brings 30 years’ experience in media and communications to the Centre from diverse roles across Australia and overseas, including in England, Africa, Hong Kong and the Galápagos Islands. She’s a Star Wars (not Trekkie) fan and dreams of a real-word Rosey robot, capable of washing and ironing… autonomously! When not sharpening her pencil, Shelley loves getting outdoors (and photographing it) with best mate, black Labrador, Josie. If you’ve got a question about the Centre or looking for story inspo, get in contact. Shelley.Thomas@qut.edu.au P: +61 7 3138 3265 M: +61 (0)416 377 444

Sponsors


Platinum Sponsor: Lockheed Martin Australia

Gold Sponsors: BHP and Monash University

Bronze Sponsors: Australian National University (ANU) and Queensland University of Technology (QUT)

For sponsor enquiries, please contact Katrina Tune via email at katrina.tune@roboticvision.org or call 0414 789 503.

 

 

 

Materials


 

Australian Centre for Robotic Vision
2 George Street Brisbane, 4001
+61 7 3138 7549