The Motion Playground tool has been developed for the AIMS5.0 project at the Eindhoven University of Technology in collaboration with Phillips. The tool can be accessed at: . Installation information can be found in the repository.
As of current Motion Playground tool has the following features:
Motion Playground provides a least-squares-based inverse kinematics solver implemented as a PyTorch module, making it easy to integrate into existing machine learning pipelines. Both the forward and inverse kinematics modules operate on kinematic models defined in axis-angle format, offering flexibility in model design. When a skeletal model is loaded from any supported file format, it is automatically converted into axis-angle representation, enabling compatibility with a variety of serialized formats.
Motion retargeting is the process of transferring skeletal animations from a source skeleton to a target skeleton. Motion Playground’s motion retargeting functionality enables accurate transfer across a wide variety of skeletons. By leveraging inverse kinematics, it can handle body parts that cannot be directly mapped due to structural differences or mismatches in the number of joints and links between the source and target skeletons.