Efficient synthesis of physically valid human motion

Abstract

Optimization is a promising way to generate new animations from a minimal amount of input data. Physically based optimization techniques, however, are difficult to scale to complex animated characters, in part because evaluating and differentiating physical quantities becomes prohibitively slow. Traditional approaches often require optimizing or constraining parameters involving joint torques; obtaining first derivatives for these parameters is generally an O(D 2 ) process, where D is the number of degrees of freedom of the character. In this paper, we describe a set of objective functions and constraints that lead to linear time analytical first derivatives. The surprising finding is that this set includes constraints on physical validity, such as ground contact constraints. Considering only constraints and objective functions that lead to linear time first derivatives results in fast per-iteration computation times and an optimization problem that appears to scale well to more complex characters. We show that qualities such as squash-and-stretch that are expected from physically based optimization result from our approach. Our animation system is particularly useful for synthesizing highly dynamic motions, and we show examples of swinging and leaping motions for characters having from 7 to 22 degrees of freedom.

Dynamic Balancing and Walking for Real-Time 3D Characters

“…A common approach to combining data driven stylistic movement with physics is to allow a controller to maintain balance while upper body movement (which is most noticeable and more intricate) is stylized [20,21]. A controller itself must rely on more than one physics technique if a range of transitional movements is to be accommodated (e.g., [1,4,10,22]). SIMBICON [10] is a notable advancement in this area in recent years and demonstrates a biped controller that utilizes a series of poses to shape motion trajectories for flexible locomotion.…”

Simulation of Human Motion Data using Short‐Horizon Model‐Predictive Control

“…These approaches are typically sensitive to gain parameters and not intuitively directed. In contrast, off-line authoring tools based on continuous optimization leverage the benefit of time to search for physical motions that are optimal according to some metric and satisfy user constraints [WK88,Coh92,PW99,FP03,SP05]. The predictive component of McSim is inspired by these off-line approaches but sacrifices global optimality for computation speed by restricting the search to a short amount of time into the future.…”

“…These contact constraints are a key component of the dynamic model used by the predictive component. Contact forces are computed using a polygonal approximation to Coulomb's model of friction [FP03]. The model is depicted in 2D in Figure 2.…”

Real-time control of physically based simulations using gentle forces

“…It has been used extensively for automatic generation of human motion [Popović and Witkin 1999;Fang and Pollard 2003;Safonova et al 2004;Sulejmanpasić and Popović 2005;Liu et al 2005]. The benefit of similar strategies has been demonstrated for control of rigidbody simulations [Popović et al 2003] and fluids [McNamara et al 2004].…”