The eleventh international conference on indoor positioning and indoor localization (IPIN2021) will be held November 29 – December 2 at Hotel Evenia Olympic Park, Lloret de Mar, Spain and online. The following oral presentation will be delivered.
In recent years, with the spread of mobile devices and IoT devices, indoor localization becomes an important technology for the basis of location information services and route guidance services.
Wi-Fi fingerprinting-based indoor localization is one of the attractive technology to realize indoor localization and actively explored in several decades.
However, Wi-Fi fingerprinting-based indoor localization has a problem that we need to acquire enough fingerprint datasets at the target environment. This task is time-consuming and labor-intensive (i.e., the data acquisition cost is high.)
Some work tries to solve this issue by deciding the data acquisition position effectively, however, the walking distance is not reduced.
To solve the cost of data acquisition, we proposed an incremental recommendation algorithm for data collection positions that reduces redundant walking during data gathering while using Bayesian optimization and multi-task learning.
With the evaluation experiment, we confirmed that the proposed algorithm can significantly reduce the data acquisition cost including walking distance and data collection time compared to the existing method.
TM2.3: WiP III, November 30, 12:00-13:00
Efficient Indoor Localization Model Construction by Sequential Recommendation of Data Gathering Position based on Bayesian Optimization
Yoshiki Omori, Masato Sugasaki and Masamichi Shimosaka