Indoor Positioning with Heterogeneous Sensors — Adaptive Virtual Reality Input Protocol for Multi-Device Integration

Heterogeneous sensors refer to sensors with different data stream types, including low-level sensors such as laser, radar, and inertial sensors, as well as high-level sensor management systems and heterogeneous data fusion algorithms. Using multi-source heterogeneous sensors for indoor pose and trajectory positioning can enhance the robustness and effectiveness of positioning algorithms, meeting the immersive requirements of virtual reality content.

Current virtual reality input methods are diversified, but various input device solutions exhibit strong coupling. This design aims to propose a plug-and-play manager and protocol for integrated input device management. This plug-and-play capability is adaptive: it can optimize methods for integrating new sensors based on user testing feedback.

This design seeks not only software protocols that enhance input device performance. The virtual reality field still contains too many traditional elements: controllers instead of glove input; stereotyped motion patterns rather than flexible and nuanced interactions. These issues are not entirely due to hardware performance limitations but also require software designers to research how virtual reality platforms present content, rather than simple migration from two-dimensional screens. Based on these platform differences, this design will appropriately propose concepts with strong conditions applicable to virtual reality that are incompatible with traditional platforms.

Since 2014, neural network methods led by deep learning have become popular in the artificial intelligence field. This design will extensively employ neural network encoder-decoders as middleware to achieve "adaptation" and "integration" functions for heterogeneous real-time sensor data.

Research Sonnet 4