16 January. 2021
We aim to develop techniques to design, control and simulate multiple characters with different behaviors that can then be applied to different environments in various application domains (computer games, movies, urban studies, etc.). Designers should have maximum levels of control of the generated simulation results with minimal user intervention. Characters should exhibit both high level behaviours such as reasoning and path planning and low-level ones such as navigation with collision avoidance and individual character animations. Additionally, animated characters can be of different types such as virtual humans, vehicles and animals. We aim to have easy and efficient multi-character design and control using notions known to animators from existing tools such as copy-pasting, painting with strokes, blending, etc. Since people are accustomed to using examples in order to describe desired requirements, we are interested in investigating novel methods that take as input exemplars of animated scenes (e.g., people behaving as one expects in a shopping street) and then given a new environment, output semantically correct behaviours for the animated characters.