DSFS: Dynamic Sensor Fusion System for Robust Localization with Diverse Sensing Information

Abstract

Autonomous mobile robots are expected to demonstrate a high degree of adaptability, enabling effective operation across diverse environments. As sensor performance is largely influenced by environmental conditions, relying solely on a single sensor for localization makes robust localization challenging. To address this issue, various studies have enhanced the localization robustness using multiple sensors with different characteristics that complement each other’s weaknesses. However, conventional studies require the design of separate fusion systems for each type and numerous sensor observations. As future developments facilitate increased cooperation with environment-fixed sensors and external agents, the types and number of sensor observations accessible to robots are expected to dynamically change depending on location and time. Therefore, a pose fusion system that adapts to such changes is required. This paper proposes a fusion system that can adapt to changes in the type and number of sensor observations. This system dynamically fuses pose information obtained from onboard sensors, environment-fixed sensors, and external agents by extending the selective fusion method, one of the existing pose fusion methods for onboard sensors. Simulation experiments confirm that our system can adapt to changes in the type and number of sensor observations and robustly localize by dynamically fusing pose information from onboard sensors, environment-fixed sensors, and external agents.