Motion planning and synthesis for virtual characters in constrained environments

Steve Tonneau 1, 2
2 MIMETIC - Analysis-Synthesis Approach for Virtual Human Simulation
UR2 - Université de Rennes 2, Inria Rennes – Bretagne Atlantique , IRISA-D6 - MEDIA ET INTERACTIONS
Abstract : With the growing complexity of virtual environments comes the need to provide virtual characters with a larger autonomy of motion. Additionally to walking, running and jumping, state of the art virtual applications require characters to climb, crawl, pull or push objects... Those tasks are characterized by the constrained environments in which they are achieved, where the risk of collision is high and motion capabilities are limited; they are also associated with important force exertion, resulting from contact creation. In this context, automatic motion synthesis is really difficult. This thesis aims at proposing an automatic method for motion synthesis in constrained environments. To achieve these goals, two research problems have been identified and studied. The first part is dedicated to the issue of generating contact postures compatible to achieve the considered tasks. We propose a new heuristic called EFORT (Extended FORce Transmission ratio). EFORT is used to evaluate the compatibility of a contact posture with the requested task. EFORT lies at the center of a new method for the real time generation of task efficient contact configurations. This generator finds its applications for arbitrary virtual characters and environment, and as such can be directly integrated within video game applications. The second part of this thesis focuses on the more global issue of computing a relevant trajectory in a constrained environment. This issue is seen as the search for a sequence of task efficient contact postures, suited for achieving the task. Consequently a new motion planner based on EFORT is proposed. Because it is one of the first to simultaneously address the complexity of the environment and task efficiency, our motion planner is a significant step towards an enhanced autonomy of motion for virtual characters.
Document type :
Complete list of metadatas

Cited literature [74 references]  Display  Hide  Download
Contributor : Abes Star <>
Submitted on : Wednesday, April 22, 2015 - 11:27:05 AM
Last modification on : Tuesday, October 1, 2019 - 3:54:14 PM
Long-term archiving on: Monday, September 14, 2015 - 12:21:46 PM


Version validated by the jury (STAR)


  • HAL Id : tel-01144630, version 1


Steve Tonneau. Motion planning and synthesis for virtual characters in constrained environments. Computer science. INSA de Rennes, 2015. English. ⟨NNT : 2015ISAR0004⟩. ⟨tel-01144630⟩



Record views


Files downloads