SCA 15: Eurographics/SIGGRAPH Symposium on Computer Animation

Permanent URI for this collection

https://diglib.eg.org/handle/10.2312/14634

Browse

Browsing SCA 15: Eurographics/SIGGRAPH Symposium on Computer Animation by Subject "control"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Hierarchical Planning and Control for Complex Motor Tasks
    (ACM Siggraph, 2015) Zimmermann, Daniel; Coros, Stelian; Ye, Yuting; Sumner, Robert W.; Gross, Markus; Florence Bertails-Descoubes and Stelian Coros and Shinjiro Sueda
    We present a planning and control framework that enables physically simulated characters to perform various types of motor tasks. To create physically-valid motion plans, our method uses a hierarchical set of simplified models. Computational resources are therefore focused where they matter most: motion plans for the immediate future are generated using higher-fidelity models, while coarser models are used to create motion plans with longer time horizons. Our framework can be used for different types of motor skills, including ones where the actions of the arms and legs must be precisely coordinated. We demonstrate controllers for tasks such as getting up from a chair, crawling onto a raised platform, or using a handrail while climbing stairs. All of the motions are simulated using a black-box physics engine from high level user commands, without requiring any motion capture data.
  • Thumbnail Image
    Item
    Learning Reduced-Order Feedback Policies for Motion Skills
    (ACM Siggraph, 2015) Ding, Kai; Liu, Libin; Panne, Michiel van de; Yin, KangKang; Florence Bertails-Descoubes and Stelian Coros and Shinjiro Sueda
    We introduce a method for learning low-dimensional linear feedback strategies for the control of physics-based animated characters around a given reference trajectory. This allows for learned low-dimensional state abstractions and action abstractions, thereby reducing the need to rely on manually designed abstractions such as the center-of-mass state or foot-placement actions. Once learned, the compact feedback structure allow simulated characters to respond to changes in the environment and changes in goals. The approach is based on policy search in the space of reduced-order linear output feedback matrices. We show that these can be used to replace or further reduce manually-designed state and action abstractions. The approach is sufficiently general to allow for the development of unconventional feedback loops, such as feedback based on ground reaction forces. Results are demonstrated for a mix of 2D and 3D systems, including tilting-platform balancing, walking, running, rolling, targeted kicks, and several types of ballhitting tasks.