ICRA 2016 Workshop
Workshop on Emerging Topological Techniques in Robotics
Stockholm, Friday, May 20th, 2016
Abstract
In the last five years, dramatic progress has been made in the field of Computational Algebraic Topology. New efficient algorithms for the computation of topological properties of data and the emergence of efficient persistent homology implementations in particular are beginning to enable novel topological approaches in a variety of fields such as Medicine, Biology and Machine Learning. Since topological properties of data are typically robust to noise and errors they are ideally suited for robotic applications with significant sensor noise. Over the past decade, robotics has benefited greatly from adopting techniques from machine learning and computational statistics to address problems of adaptation and model learning. Recently developed tools for computational algebraic topology and discrete Morse theory represent an opportunity to add to this toolbox, especially in the area of unsupervised learning and categorisation, e.g., discovering structured latent spaces to describe articulated full body motion. This workshop seeks to popularize the use of recent topological techniques in robotics and will encourage collaboration between the computational geometry and robotics communities to promote and discuss the use of topological methods in Robotics. The workshop will focus on three sub-themes: 1) Introduction, Algorithms and Tools for Topological Methods, 2) Homotopy Aware Motion Planning and 3) Sensor Networks and Related Emerging Applications.
Workshop Poster Submission - deadline 8th April 2016
We highly encourage submission of posters for the workshop. If you would like to present recent and ongoing work on topological approaches for robotics, please submit a 1-4 page abstract outlining your research. We are planning to have a short poster highlight session as well as a dedicated poster session during the workshop, providing a unique opportunity to discuss your work in this area with colleagues.
Schedule
13:30 - 13:45 | Welcome and Introduction |
13:45 - 14:15 | "Topological motion planning" - Subhrajit Bhattacharya, UPenn |
14:15 - 14:45 | "Persistence and topological data analysis for robotics" - Florian T. Pokorny, KTH |
14:45 - 15:20 | Poster Spotlight |
15:20 - 15:45 | Coffee Break |
15:45 - 16:15 | Uses of persistence for interpreting coarse instructions - Subramanian Ramamoorthy, University of Edinburgh | 16:15 - 16:45 | "Coarse State Estimation in GPS-Denied Environments via Homological Sensing" - Jason Derenick, Exyn Technologies |
16:45 - 17:15 | "Search-based Motion Planning with Topology-based Heuristic Functions" - Maxim Likhachev, CMU |
17:15 - 17:45 | "Topological Abstraction and Planning for Pursuit-Evasion Problem, Alberto Speranzon, Honeywell |
17:45 - 18:00 | Closing remarks and discussion |
Posters
Asymptotically Optimal Sampling-Based Topological Motion Planning Algorithms and its Application to Stochastic Optimal Control, Jung-Su Ha and Han-Lim Choi |
Topological mapping using a heterogeneous robotic swarm, Ragesh K. Ramachandran and Spring Berman |
Some progress toward tethered pairs, Reza H. Teshnizi and Dylan A. Shell |
A Homological Approach for Loop Detection Using Three-dimensional Normal Distributions Transform, Ji W. Kim , Sang Hoon Ji , Sujeong You , Hong Seok Kim , and Beom H. Lee |
3D Point Cloud Segmentation Using Topological Persistence, William J. Beksi and Nikolas Papanikolopoulos |
Clustering-Based Robot Navigation and Control, Omur Arslan, Dan P. Gulranik, and Daniel E. Koditschek |
Organizers
Subhrajit Bhattacharya (UPenn)
Vijay Kumar (UPenn)
Florian T. Pokorny (KTH)
Subramanian Ramamoorthy (Edinburgh University)