Image-Assisted Beamforming for Accurate User Localization

The patent combines audio data from microphone arrays with image data from cameras to perform joint user position estimation. This multi-sensor fusion approach resolves the technical contradiction by merging complementary data sources: audio provides directional information while images provide visual confirmation of user location, together achieving high reliability in beamforming target selection without merely adding complexity but creating synergistic value.

The patent introduces an intermediary processing layer that fuses audio and image data through confidence-based user position estimation. This intermediary layer processes multiple data sources, reconciles their different characteristics, and produces a unified user position estimate that guides beamforming, thereby resolving the contradiction between using simple audio-only sensing and achieving accurate beamforming through complex multi-sensor systems.

The patent implements partial action by using multiple sensors only when necessary - specifically when audio-based user position estimation confidence is below a threshold. When audio confidence is sufficient, the system uses audio alone, avoiding the computational overhead of processing image data. This resolves the contradiction by applying multi-sensor processing selectively rather than continuously, achieving high measurement precision only when needed while conserving computational energy during routine operations.

The patent dynamically changes the processing parameters based on audio confidence levels. When confidence is high, the system processes only audio data (lower computational load). When confidence is low, it activates image processing (higher computational load) to improve position estimation accuracy. This parameter-based adaptive approach resolves the contradiction between measurement precision and energy consumption by adjusting the level of processing based on real-time conditions.

The patent applies partial action by computing beamforming only in the direction of the estimated user position rather than computing multiple beams for all possible directions. By using confidence-based position estimation from fused audio-image data, the system achieves sufficient adaptability to handle uncertain positions through accurate single-direction estimation, thereby avoiding the computational waste of calculating unnecessary multiple beams while maintaining versatility.

The patent substitutes the mechanical approach of computing multiple beams simultaneously with a more efficient method: using sensor fusion to accurately estimate user position and compute a single targeted beam. This replaces the brute-force multi-directional computation with an intelligent, data-driven single-direction approach, resolving the contradiction between adaptability to uncertain positions and computational efficiency by using information fusion rather than exhaustive computation.