Support for the indoor positioning of personal devices has been attracting a lot of interest in the sensor networks community and from robotics researchers. The authors developed a knowledge base to better account for different sensors. The paper presents a model and a set of experiments. The state of the art in localization and tracking integrates sensor data (from Wi-Fi and wireless sensor networks) using probabilistic methods such as particle filters (PFs). Since those approaches strongly depend on sensor and optimization functions, the authors propose a declarative programming approach that allows for more flexible and adaptable criteria for the integration.

After the introductory section, the second section describes the sensors used, the simulation scenario for the experiments, and the data acquisition tools used to gather real data in an indoor laboratory environment.

Section 3 defines the localization and tracking problems, together with the metrics, including the best proximity, best support, best move, and best coherence metrics, and these are applied to a set of experiments for tracking humans. The best location at each time step depends on the location selected at the previous time step. Localization is expressed as the coordinates of the 2D cell where the mobile agent is localized.

In the fourth section, the knowledge base for indoor positioning is defined in the answer set programming (ASP) logic framework. Localization derives from the verification of properties of the candidate solutions that arise from sensor data. The best solution is chosen using ASP’s generate-and-test method.

Section 5 reports on the experiments, which use simulated and real data. Different parameters are computed using a PF and ASP over a large number of time steps and noise hypotheses. The results for ASP are only slightly more accurate than the results for the PF. However, for robustness, ASP shows a clear advantage.

The paper is readable and well organized. It achieves its main goal: to demonstrate the superiority of hybrid systems that are able to support localization methods with semantic information. Future work could improve on the limited experimental scenarios (due to space and sensor availability) and move toward real-time applications.