Oct72009 < LCI

---++ Minutes:

   * 2-page abstract due for this year. We'll develop it in the directory:
      * DOC/abstract inside SVN. Dave will start it and email around [DM]
   * Qualification video - take it sometime that's convenient

   * Start planning itinerary
      * 12 possibly interested {TH, AG, MM, CG, PF, MB, SH, DM, PV, SM, WW, TS}
      * SH to email the 4 supervisors, who will figure it out for us :) {JL, DL, BW, AM}
      * For budgeting reasons, the costs are roughly:
         * >= $380 US round-trip for flights
         * ~ $55/night per room at the Monte Carlo (cheaper elsewhere)
         * ~ $2000 for robot shipping
         * Potentially some conference registrations ($400.00 for students)

   * Catherine update on shipping
      * Cost for shipping is about the same as cost to rent a car and drive
      * MM might be ok with driving his own car. Wonder how much UBC/SRVC would compensate him for putting 6K on his new Jetta
      * Debate on driving vs shipping on several axes:
         * Shipping:
            * Pros: Doesnt take our time, if we invest in a crate (~$800), we always have it later
            * Cons: Potential for damage, extra downtime for robot with packaging
         * Driving:
            * Pros: We keep the robot in our possession, saves cost on flights 
            * Cons: Takes a lot of our time
      * Major issue seems to be how long the shipping downtime would be. CG to check out.
      * Also would like to know if it's possible to rent a truck with unlimited km. SH to check out.
      * Dont forget that loading the robot into anything without a ramp is difficult.
      * Need to decide soon. 
   * Theakston and Lapinkulta are self-admined and can be used for development. Log in with lciuser... and sudo adduser to make your own account.
   * Finally, the list of completed and in progress tasks.


---++ Completed components:

   * Porting of basic drivers for: bumblebee, cannon and powerbot
   * Tower design
   * gmapping
   * Tilting laser drivers
   * Robot coordinate transform code
   * Network configuration and development environment
      * Robot router setup
      * Setup self-administered PC's
      * ROS instructions
   
---++Current in-progress task list:

   * Capture data from robot for testing
   * Basic robot functions based on ROS. (with aim to perform a preliminary test run of navigation and mapping) [MM and DM]
      * WG nav stack
      * Tower upgrade:
         * Order material for building a new laser/camera mount and assemble same.
      * Saliency maps and visual attention 
         * Basic saliency map computation [DM]
         * Stereo + saliency combined to identify interesting regions [PV]
   * High-level control functionality such as planning
      * Random walk behavior
      * 3 main high-level planners:
         * Exploring frontiers
         * Find tables [PV]
         * Space coverage
         * Look back at objects
      * Top level state machine to choose between above planners
      * Choice of "where-to-look" aka attention system
   * Recognition framework (James module directly or something built upon that) [AG and CG]
      * Combining results from different types of detectors (different algorithms)
      * Combining results from various viewpoints
          * We'll meet on the previous two topics tomorrow
      * Collect data for 5 "given" object classes once they're published
      * Test data interface
      * Felzenswalb detector
         * MB profiled Kenji's python implementation - most of the time in convolution - promising
         * Will investigate cuda'ing pieces
      * Helmer detector
         * Using point cloud, 
      * Mccann
      * Training data interface and additional parameters 
   * Cuda on fraser [MB, WW and TH]
      * Need to get the code compiling
      * GPUSift
      * FastHOG
   * Web grabbing module [PF and CG]
      * Add additional sources of info
      * Investigate filtering techniques
      * Integrate output data format with classification
   * Speed-up of Felzenswalb training [MB]
      * Initial investigation to verify this is a doable task (profiling current code, ensuring good performance on web data, investigation of potential speedups such as GPU feature extraction and SVM learning)

---++ Future tasks pending completion of others:

   * Use of 3D models in recognition
   * Use of 3D information and context in attention system
   * Real time result reporting
   * Feeding back classification results to robot planner
   * Investigate new cameras which might be faster than the Cannon
   * Prioritizing computation done by classifiers towards images which look really promising to the attention system, and based on the classes which have already been recognized.
This topic: LCI > WebHome > TheSemanticRobotVisionChallenge > Oct72009
Topic revision: r4 - 2009-10-14 - DavidMeger