The 2009 International Conference on Robotics and Automation just finished in Kobe, Japan. Notable in addition to the technical content of the conference were the measures being taken regarding swine flu.
Among the hundreds to papers that were presented there were a couple that discussed recent developments regarding automated flight be quad-rotor helicopters. Wolfram Burgard's group won the best paper prize for work on doing automated mapping and localization (SLAM) with such a helicopter. Our friend and former McGill student Nick Roy also won the early career award for his achievements, which include work on such helicopters (mentioned on this blog a few weeks ago).
Junaed gave a presentation of our work using "spatio-chromatic" trackers to follow human dives with our underwater robot. This work uses a very large set of possible image filters with both spatial and color tunings to detect the person. The combination of filters to use is practice is optimized using the Adaboost machine learning algorithm.
Philippe gave a talk on our work using simple contact dynamics to measure the surface texture of the ground the robot is walking on, and classify and identify it using unsupervised learning.
Yiannis presented work with Dave and Dimitri on using a combination of Extended Kalman Filters and Monte Carlo Markov Chain methods for state estimation. The work is applied to computing node positions for a sensor network. In that work, the hybrid MCMC/EKF combines advantages of each individual algorithm.
