Final Activity Report Summary - RAGA (Robotics Algorithms for Computer Animation)
The main objective of the RAGA project was to transfer techniques developed and used by the robotics field in the computer animation domain. Our work focused on the motion planning problem, i.e. on how to move in a given environment while avoiding obstacles.
We considered the specific problem of virtual crowds. Virtual crowds were made of numerous digital actors, up to tens of thousands, living in a virtual world simulated on a computer. Considering real-time crowds was a challenge, because a single computer should simulate and render thousands of elements interactively with the user or the spectator.
During the RAGA project, we developed tools allowing a user:
1. to design a virtual population for its favourite virtual world by attributing goals to each individual. This process was made easy and intuitive via a small set of high level directives;
2. to simulate the crowd thus designed in real-time; and
3. to render them on desktop PC.
Computation times required by each previous element were small enough to allow for an interactive use. As a result, our solutions fit entertainment applications such as video-games.
The solution was based on an analysis of the environment, allowing for pre-computing a data structure that was used during the design and simulation stages. The solution was also scalable, in order to be able to consider very large crowds and scenes. The computation resources were distributed in space and time, so that the quality and precision of the simulation were best where the attention of the spectator was focused.
We considered the specific problem of virtual crowds. Virtual crowds were made of numerous digital actors, up to tens of thousands, living in a virtual world simulated on a computer. Considering real-time crowds was a challenge, because a single computer should simulate and render thousands of elements interactively with the user or the spectator.
During the RAGA project, we developed tools allowing a user:
1. to design a virtual population for its favourite virtual world by attributing goals to each individual. This process was made easy and intuitive via a small set of high level directives;
2. to simulate the crowd thus designed in real-time; and
3. to render them on desktop PC.
Computation times required by each previous element were small enough to allow for an interactive use. As a result, our solutions fit entertainment applications such as video-games.
The solution was based on an analysis of the environment, allowing for pre-computing a data structure that was used during the design and simulation stages. The solution was also scalable, in order to be able to consider very large crowds and scenes. The computation resources were distributed in space and time, so that the quality and precision of the simulation were best where the attention of the spectator was focused.