Skip to Main content Skip to Navigation

Navigation intelligente dans des environnements virtuels

Abstract : Virtual reality (VR) has experienced fast development thanks to technologies from 3D computer games and affordable head-mounted devices (HMDs), making it possible to interact with virtual environments more easily. During the interaction process, the user is at the center of the application, and therefore, it is paramount to understand the human-related factors affecting user experience and satisfaction. These factors are crucial for designing virtual navigation as natural as users usually conduct subconsciously in the physical world.However, navigation techniques in virtual environments might be cognitively demanding and induce cybersickness, leading to the rejection of VR technologies. Various human-related factors can influence user experience, especially the level of cybersickness during immersion in VR applications. As these effects are individually different, its correct evaluation is a premise for designing different navigation techniques. To use these individual differences for adapting the navigation dynamics, we successfully introduced a fuzzy logic model to analyze human characteristics related to cybersickness and output a personalized cybersickness indicator. On top of the current evaluation approaches, we demonstrated how to measure cybersickness and cognitive workload based on deep learning (LSTM autoencoder) and TOPSIS. The proposed evaluation method can be regarded as an improvement for the current evaluation approaches. The results suggest that deep learning represents an interesting innovative alternative to measure cybersickness based on behavioral assets such as posture sway, and TOPSIS can improve the measuring accuracy for the cognitive workload.The progress in unlocking the potential of VR also relies on our ability to develop navigation techniques that can mitigate cybersickness efficiently. We introduced adaptive navigation in the sense of personalization and designed four different original navigation techniques considering human-related factors. First, we proposed semiautomatic navigation to smoothen navigation trajectories during user displacement. Second, knowing the importance of navigation speed profiles on user experience, we designed a speed protector that can minimize the total jerk when the user navigates in virtual environments. Third, the fuzzy logic model introduced to compute a personalized cybersickness indicator was used to propose personalized navigation dynamics. Last, based on the PID control theory and neural networks, we designed an online adaptive strategy to adapt the navigation speed based on physiological assets. For each navigation technique, user experiments were performed to validate its performance in improving user experience, and the results manifest a significant reduction of cybersickness severity.
Document type :
Complete list of metadata
Contributor : ABES STAR :  Contact
Submitted on : Monday, May 30, 2022 - 3:12:35 PM
Last modification on : Friday, August 5, 2022 - 2:54:01 PM
Long-term archiving on: : Wednesday, August 31, 2022 - 6:51:56 PM


Version validated by the jury (STAR)


  • HAL Id : tel-03681759, version 1


Yuyang Wang. Navigation intelligente dans des environnements virtuels. Traitement du signal et de l'image [eess.SP]. HESAM Université, 2021. Français. ⟨NNT : 2021HESAE043⟩. ⟨tel-03681759⟩



Record views


Files downloads