Continuing the Siggraph 2007 roundup for character animation (see this previous post to catch up). The remaining papers take an approach leaning towards physical simulation rather than pure motion capture.
One paper from Seoul National University Movement Research Lab presents an biped behaviors simulated from human motion data. Once again, this project has two innovations:
An optimization technique that can transform mo-cap data into a balancing motion that can be physically simulated.
A controller learning algorithm to create robust dynamic controllers from the raw data.
The advantage of this approach is being based on a physically accurate simulation, so it’s possible to generate new animations dynamically. It also combines the realism of motion capture by imitating the input in a physically correct way.
View or download the movie (MOV, 120 Mb).
Here’s the abstract:
Physically based simulation of human motions is an important issue in the context of computer animation, robotics and biomechanics. We present a new technique for allowing our physically-simulated planar biped characters to imitate human behaviors. Our contribution is twofold. We developed an optimization method that transforms any (either motion-captured or kinematically synthesized) biped motion into a physically-feasible, balance-maintaining simulated motion. Our optimization method allows us to collect a rich set of training data that contains stylistic, personality-rich human behaviors.
Our controller learning algorithm facilitates the creation and composition of robust dynamic controllers that are learned from training data. We demonstrate a planar articulated character that is dynamically simulated in real time, equipped with an integrated repertoire of motor skills, and controlled interactively to perform desired motions.
Download the paper (PDF, 0.8 Mb):
Simulating Biped Behaviors from Human Motion Data Kwang Won Sok, Manmyung Kim, Jehee Lee Siggraph 2007
Here’s a quick assessment of the technology based on how easy it would be to use for upcoming games.
- Applicability to games: 2/10
- Sadly, this technology only works in a 2D plane for the moment, so its application is limited. If it supported 3D, this score would go up to by quite a few points, as it would compete in quality with Natural Motion’s animation middleware.
- Usefulness for character AI: 3/10
- For the same reasons, applicability to 3D AI characters is limited. However, conceptually it’s very useful to have a physical controller at the base of the animation; it becomes much easier to synthesize animations dynamically.
- Simplicity to implement: 3/10
- Physically-based simulation is not easy to get right in general. This implementation requires good knowledge of kinematics and general optimization theory.
Capturing Motion for the Walk Cycle
How do you think this kind of technology can be useful to game AI?