Multi-channel haptic feedback, e.g., vibrotactile, is essential to realize truly immersive extended-reality (XR) experiences. Such XR applications increasingly demand offline-capable solutions, where multi-channel haptic feedback is measured offline and delivered in delay-sensitive applications, such as virtual training simulators and museum exhibits. A key challenge in enabling such experiences is compressing and delivering high-perceptual-quality multi-channel vibrotactile signals under limited available bandwidth. The state-of-the-art vibrotactile codec for the pre-measured multi-channel vibrotactile signals uses psycho-haptic models to remove the perceptual redundancy to realize a compact representation. However, the coding efficiency remains limited under bandwidth constraints. We propose a hybrid offline–online framework that leverages both offline coding and online transmission phases for delivering high-perceptual-quality multi-channel vibrotactile. In the offline coding phase, the pre-measured multi-channel vibrotactile signals are distilled into a compact implicit neural representation (INR) considering the psychohaptic model. In the online transmission phase, only lightweight residuals are compressed and transmitted to adaptively refine the vibrotactile signals under the available bandwidth. This hybrid approach leverages offline perception-aware INR to generate compact side information under a storage budget, while enabling online residual transmission to adaptively refine vibrotactile quality based on instantaneous bandwidth. Experiments on an open multi-channel vibrotactile dataset demonstrate that the proposed scheme improves perceptual quality compared to existing multi-channel vibrotactile coding schemes with the same available bandwidth. Additionally, the proposed human-tactile perception-aware INR can be considered better side information for improving the quality of online residual transmission.
Ozawa et al. (Thu,) studied this question.