Hybrid indoor navigation with BLE RSSI probabilistic homing and vision-assisted guidance
TL;DR
This paper presents a low-cost hybrid indoor navigation architecture that combines three complementary navigation modalities within a single distributed embedded platform and demonstrates successful obstacle recovery, state-timeout handling, and stable REACHED-state operation under noisy indoor conditions without relying on SLAM or expensive LiDAR infrastructure.
Abstract
Autonomous indoor navigation remains a significant challenge because Global Positioning System (GPS) signals are often unavailable or unreliable within indoor environments. Although LiDAR-based Simultaneous Localization and Mapping (SLAM) techniques provide accurate localization and mapping capabilities, they require expensive hardware and substantial computational resources, making them unsuitable for low-cost and resource-constrained platforms. This paper presents a low-cost hybrid indoor navigation architecture that combines three complementary navigation modalities within a single distributed embedded platform. The principal technical contribution is a BLE RSSI-based probabilistic directional homing system in which raw signal volatility is tamed through median filtering and exponential moving average (EMA) smoothing, while navigation decisions are driven by a finite state machine (FSM) that responds to RSSI trend classification rather than instantaneous signal magnitude. An AMBIGUOUS recovery state enables directional rescanning whenever obstacle blockage, sustained plateau conditions, or unstable RSSI behaviour makes the current heading unreliable. A supporting precision module employs overhead ArUco marker localization and differential steering control for localized trajectory refinement near the target region. Experimental BLE navigation trials demonstrated successful obstacle recovery, state-timeout handling, and stable REACHED-state operation under noisy indoor conditions without relying on SLAM or expensive LiDAR infrastructure.