ShelfAware: Real-Time Semantic Localization in Quasi-Static Environments with Low-Cost Sensors

ShelfAware is a semantic particle filter system that enables robust indoor localization in dynamic, cluttered environments using low-cost vision sensors. By treating scene semantics as statistical evidence rather than fixed landmarks, the technology achieves 97% global localization success in retail settings and outperforms existing geometric and semantic baselines.

ShelfAware addresses a critical limitation in robotics and indoor navigation: the failure of standard vision-based localization in environments where geometry repeats but semantics shift constantly. Retail stores, warehouses, and similar quasi-static spaces present unique challenges—shelves look identical, objects move frequently, and clutter creates perceptual noise that breaks traditional SLAM and landmark-based approaches. This research tackles the problem by shifting from treating objects as fixed quantity landmarks to modeling them as probabilistic semantic categories, enabling the system to leverage distributed semantic evidence through a Monte Carlo Localization framework.

The technical innovation centers on inverse semantic proposals paired with precomputed viewpoint banks, allowing fast hypothesis generation without expensive hardware. Evaluation across both controlled mock retail environments (97% success rate) and operational 3,500 sq. ft. grocery stores demonstrates real-world viability. The system maintains high tracking accuracy (66%) even under dynamic occlusion and cart movement, conditions that typically degrade localization performance.

For the robotics and autonomous systems industry, ShelfAware represents meaningful progress toward infrastructure-free deployment in dynamic commercial spaces. Mobile robots, assistive robots, and autonomous inventory systems require reliable localization without expensive sensors or pre-installed infrastructure. This work bridges that gap by proving semantic reasoning can replace geometric certainty in challenging environments. The perception-agnostic approach—demonstrated across two distinct domains—suggests broad applicability across retail, logistics, and facility management sectors.

• →ShelfAware achieves 97% global localization success in retail environments by modeling object semantics as probabilistic distributions rather than fixed landmarks.
• →The system uses inverse semantic proposals with precomputed viewpoint banks to generate targeted Monte Carlo hypotheses on low-cost vision-only hardware.
• →Real-world testing in operational grocery stores shows significant outperformance over both geometric and fixed-quantity semantic baselines.
• →The technology maintains 66% tracking success under dynamic occlusion and moving obstacles, addressing practical deployment challenges.
• →Infrastructure-free operation makes ShelfAware applicable to mobile robots, assistive robots, and autonomous systems in commercial environments.