Autonomous Character Navigation in Complex Environments

Readings:

1) Precomputed Search Trees: Planning for Interactive Goal Driven Animation: M.Lau, J.Kuffner
2) Construction and optimal search of interpolated motion graphs: A.Safonova, J.K.Hodgins
3) Planning biped locomotion using motion capture data and probabilistic roadmaps: M.Choi, J.Lee, S.Y.Shin
4) Fast and accurate goal-directed motion synthesis for crowds: M.Sung, L.Kovar, M.Gleicher
5) Learning to Move Autonomously in a Hostile World: L.Ikemoto, O.Arikan, D.Forsyth
6) Near-optimal Character Animation with Continuous Control: A.Treuille, Y.Lee, Z.Popovic

November 7 Update:

1) Fixed avatar orientation issues in RVO by performing look-ahead over the duration of a walk cycle
2) Animated skeletal characters with a simple, single walk cycle and displayed them in callisto (default viewer)
3) Gathered motion capture data for locomotion (walk and run cycles) (acknowledgement: Taesoo Kwon, KAIST)
4) Built Horde3D and wrote a small snippet of code to read in output from RVO and animate the character from the Chicago sample demo

Goals for project

1) Segment and label the motion capture dataset to create a FSM of stop, walk and run motions performed at varying speeds (approach similar to Online Locomotion Synthesis, Kwon, T and Shin, S.Y, SCA 2005)
2) First step is to animate the skeletal character with mocap data inside callisto and then move onto rendering in Horde3D (or maybe offline rendering if feasible)
3) Replace the character from the Chicago sample with a couple of other rigged characters (ask Sean or others for help here) to introduce variability and create demos