Section: New Results

Outline

In 2017, MimeTIC has maintained his activity in motion analysis, modelling and simulation. In motion analysis, we focused our efforts on three major points: 1) being able to simplify the calibration and simulation of customized musculoskeletal models of the subjects, 2) explore how visual perception act on collision avoidance in pedestrian locomotion with an extension to group behavior, and 3) adapt accurate analysis in real condition (industrial or clinical contexts) where measurement inaccuracies and easy-to-use constraints make it difficult to directly apply methods used in laboratories.

From a long time, MimeTIC has been promoting the idea of using Virtual Reality to train human performance. On the one hand, it leads to an efficient tradeoff between high control and naturalness of the situation. On the other hand, it raises several fundamental questions about the automatic evaluation of the performance of the user, and the transfer of the skills trained in VR to real practice. In 2017, we explored these two questions by 1) developping new automatic methods for users' performance recognition and evaluation, and 2) biofidelity of mass manipulation in VR using haptic interfaces.

In virtual cinematography, we applied the analysis/synthesis approach to extract and simulate film styles and narration. We also extended our previously defiend Toric Space for camera placement to drone toric space to control a group of drones filming the action of an actor to ensure the coverture of cinematographic distinct viewpoints.