I have been appointed associate editor for a new IEEE journal: The Robotics & Automation Letters Journal aims at rapid publication within 6 months of submission, and has an interesting ICRA track option. Papers submitted during a certain time window can be presented at ICRA when accepted for the journal. I am looking forward to this exciting new role in the research community.
We ran two successful workshops on Visual Place Recognition in Changing Environments (VPRiCE) at two major robotics and computer vision conferences, ICRA and CVPR. At ICRA in Seattle around 130 people followed the invited talks and paper presentations in a large ballroom (see above). Two weeks later during the CVPR edition, approximately 40 people came by for talks and poster presentations. It was great to interact with the authors and of course the invited speakers Josef Sivic and John Leonard at CVPR as well as David Cox and Chi Hay Tong at ICRA. Thanks everybody for contributing!
I co-organized these workshops with Michael Milford and Peter Corke from QUT, Brisbane. For the CVPR edition we got support from Torsten Sattler from ETH Zürich.
I am part of the team that won the best paper award at the workshop on Vision in Vehicle Technology at CVPR in Boston. This paper – entitled „Sequence Searching With Deep-Learnt Depth for Condition- and Viewpoint-Invariant Route-Based Place Recognition“ – was lead by Michael Milford with support from a great team from QUT and the University of Adelaide.
Nvidia was so kind to sponsor a Titan X GPU to complement the certificate.
Both of our papers have been accepted for the Scene Understanding Workshop at CVPR in Boston.
One paper is entitled „Continuous Factor Graphs for Holistic Scene Understanding„, the other was headed by Sean McMahon and asks „How Good are EdgeBoxes, Really?“.
I am currently in Zürich, visiting the Computer Vision and Geometry lab lead by at ETH. Marc is an international partner investigator in our and an expert in computer vision and 3d reconstruction. Thanks for hosting me for three weeks in this beautiful city.
While I am here, I also got a tour through the Autonomous Systems Lab headed by Roland Siegwart.
I am happy to be invited to the Seminar on Vision for Autonomous Vehicles and Probes at Schloss Dagstuhl (near Saarbrücken) in November. This is a rather renowned international seminar series and participation is by invitation only. I am looking forward to it!
Our paper „Evaluation of Features for Leaf Classification in Challenging Conditions“ has been accepted for publication at the WACV conference in January.
In this work we analyzed the performance of different features for the task of leaf classification (given a leaf, identify the plant species). This work is part of our research in agricultural robotics and possible applications are the identification of different weed and crop species for a smarter weed management methods that use less or no herbicides and are therefore much more environment-friendly.
We submitted a paper on place categorization and semantic mapping using convolutional networks to ICRA. Check out the website for a number of videos demonstrating the approach. A second paper we submitted analyses the performance of ConvNet features for place recognition in changing environments.
I am very happy to announce that I officially started working with the Australian Centre for Robotic Vision (ACRV) today. I will be working closely with Michael Milford and Ben Upcroft on the topic of robust vision. We will continue to work on robust visual representations for place recognition under extreme appearance changes and will address this problem by utilising and extending various deep learning techniques.
I am looking forward to working with all the great researchers from the four involved Australian universities in Brisbane, Adelaide, Melbourne, and Canberra! I also hope to visit and work with some of the international partners of the centre in London, Oxford, Zürich, Atlanta and Rennes.
I visited the i-SAIRAS conference at McGill university in Montreal. This was a nice opportunity to catch up with my former Professor from Chemnitz University and other space robotics researchers from around the world. My colleagues from Chemnitz and I also published a paper at the conference about our experiences in the DLR SpaceBot Cup.
On the last day at the conference we got a guided tour at the Canadian Space Agency. We were allowed to take photos outside at their Mars yard, so here are some impressions. An ensemble of cool robots, unfortunately none demonstrated any autonomous behavior, they were all remotely controlled from on-site or the control room. Still a very nice show.
Inside they had more robots and a to-scale model of the Canadarm that is operating at the space station. Pretty impressive, the thing is huge! But I was surprised to hear that there are no safety measures with respect to collision avoidance, the astronauts could drive the arm into the space station if they are not careful.
The Workshop on Visual Place Recognition in Changing Environments I co-organized with Michael Milford and Peter Corke at ICRA 2013 in Hong Kong has been a good success. We had 9 contributed papers, a tutorial given by Peter, and invited talks by Michael and Paul Newman. The papers, and some of the presentations and posters are online here.
We are thinking of organizing another edition of this workshop at ICRA next year, so stay tuned.
In joint work with Chris McCool and Ben Upcroft, we participated in the LifeCLEF Plant Recognition Task and applied generic CNN features (using Decaf as a feature detector) in combination with extremely randomized trees to classify 500 different plant species. We ranked 4th (of 10) in the team ranking, which is not bad given this was our first shot at this problem. We will wrap this work up by submitting a paper to the LifeClef workshop.
I started working with the Robotics and Autonomous Systems group at Queensland University of Technology in Brisbane. I will join the Agricultural Robotics Programme, working on the visual identification of different plant species.