RSS 2016 Workshop on Visual Place Recognition: What is it Good For?

This half-day workshop discusses novel concepts and ideas for robust vision‐based place recognition in severely changing environments as well as discussing the extent to which place recognition is useful, or even required for robots. Such changes – induced e.g. by the time of day, weather or seasonal effects as well as human activity – are a ubiquitous challenge for all autonomous systems aiming at long‐term operations in both indoor and outdoor settings, and are highly relevant to visual neuroscience.

This year’s edition of the workshop puts a focus on contributions that show, discuss, and evaluate visual place recognition working in real robotic systems, in concert with mapping or SLAM, or embedded into a real-world application. The workshop features invited talks and a number of contributed talks discuss the newest developments, concepts and ideas in the areas covered by the workshop topics.


The workshop will take place in conjunction with the Robotics Science and Systems (RSS) conference at the University of Michigan on June 19th 2016.

Invited Speakers

Important Dates

  • Extended Deadline: May 22: (anywhere on earth) Submission of contributed papers 
  • Submitted papers will be reviewed quickly and the authors will be informed in a timely manner.
  • Sunday 19 June, afternoon : Workshop held at RSS (University of Michigan, Ann Arbor, USA)

Program and Contributed Papers

TimeProgram Item

Invited Talk: Jose Neira: The Chimera of Robust Place Recognition

In this talk I will discuss the history of the place recognition, or loop closing problem in SLAM. I will also present some of the most recent algorithms and results of our group in this field, and I will also explain why hoping for a fool-proof place recognition algorithm is a chimera, and SLAM systems should instead accommodate for possible failures in place recognition.


Niko Sünderhauf: Deep Learning for Place Recognition What to do if Place Recognition Fails


Invited Talk: Jose M.M. Montiel: Visual Place recognition in ORBSLAM

The talk focuses on the ORBSLAM a bag of words place recognizer built on DBoW2 with ORB. ORB are binary features invariant to rotation and scale, resulting in a very fast recognizer with good invariance to viewpoint. The system has been designed targeting a tight integration of the place recognition at the core the visual SLAM architecture. Thanks to this tight integration the system achieves a reliable relocation and loop closure capabilities that warrants the full system robust performance in realistic operational environments.

15:30Coffee Break
16:00Fei Han et al. – Life-Long Place Recognition by Shared Representative Appearance Learning

Muneeb Shahid, Tayyab Naseer, Wolfram BurgardDTLC: Deeply Trained Loop Closure Detections for Lifelong Visual SLAM


Charbel Azzi et al. – Global Descriptors Reduce the Image-Based Localization Search Space

16:45Stephanie Lowry and Henrik Andreasson – Visual place recognition techniques for pose estimation in changing environments
17:00Concluding Remarks and Discussion

Topics of Interest 

This year’s edition of the workshop puts a focus on contributions that show, discuss, and evaluate visual place recognition working in real robotic systems, in concert with mapping or SLAM, or embedded into a real-world application.Topics of interest to this workshop include, but are not necessarily limited to:

  • Novel techniques for change-invariant whole image matching
  • Approaches for invariant image feature learning 
  • Learning, modelling, and predicting systematic or repeating appearance changes over time
  • Learning stable / non-changing environmental features
  • Exploiting semantic information for long-term place recognition
  • Novel concepts of incorporating uncertain place recognition in SLAM / navigation systems
  • Standardized benchmarks and long-term datasets in changing environments

We explicitly encourage the submission of papers describing work in progress, preliminary results or novel concepts.


The workshop is organized by Dr Niko Sünderhauf, Assoc. Professor Michael Milford, Assoc. Professor Ben Upcroft, and Professor Peter Corke who are with the Australian Centre for Robotic Vision at QUT in Brisbane, Australia and Dr Peer Neubert from Chemnitz University of Technology, Germany.

Contact details:

Dr Niko Sünderhauf
Australian Centre for Robotic Vision
School of Electrical Engineering and Computer Science
Queensland University of Technology
Brisbane QLD 4000, Australia