Fusion of measurement data from impulse-radar sensors and depth sensors when applied for monitoring of elderly and disabled persons
Research project financed by the National Science Centre (Poland)
Grant no. 2017/25/N/ST7/00411
2018–2021
Publications
About the project
It is expected that the share of European population reaching the age of 65 years or more will increase by at least 50% during the next 30 years. The problem of organised care over elderly people is, therefore, of growing importance. Hence the demand for research on new technologies that could be employed in care services for such people. Its primary objective is to examine the applicability of various sensors for non-invasive monitoring of the movements and vital bodily functions, such as heartbeat or breathing rhythm, of elderly persons in their home environment. Moreover, the systems for monitoring of elderly and disabled persons are expected to detect dangerous events, such as falls, but also to predict them on the basis of acquired data, and therefore contribute to the prevention of such events: the analysis of gait, itinerary and timing of activities of monitored persons is intended for this purpose.
The project was aimed at investigation of the possibility of increasing the reliability of monitoring of elderly persons by means of the combination of non-intrusive sensors of two different types – viz. impulse-radar sensors and a depth sensor – followed by adequate fusion of the data provided by these sensors.
Within the Project, a library of algorithms for fusion of measurement data acquired by means of radar sensors and depth sensors was developed, and the performance of these algorithms when applied for estimation of the position of a monitored person, and of other health-related quantities, was assessed.
If the monitoring systems based on only one type of sensors are considered, the results of the experiments seem to speak in favor of the monitoring system based on the depth sensor, since the position estimates obtained by means of this sensor are significantly more accurate. However, the depth-sensor-based system has also some disadvantages if compared to the radar-sensor-based system:
- The field of view of the depth sensor is limited to a single room; even if multiple depth sensors are used for monitoring of several rooms in a household, they are still unable to track a person in certain hidden places, e.g. behind furniture; on the other hand, a pair of radar sensors enables through-the-wall monitoring, and therefore the covering of a larger area without the need of installing multiple pairs of sensors.
- Even though the acquisition of depth data infringes the privacy of the monitored persons less than vision-based solutions, it may be less acceptable for them than the acquisition of radar data – which do not provide silhouette information.
- It has been observed in the completed experiments that the fraction of missing data may be very high in case of the depth sensor exposed to strong sunlight – which does not affect the radar sensors at all.
If the data fusion in considered, the numerical experiments have demonstrated that the fusion of data from sensors of both types alleviates some important problems which occur if only single-type sensors are used for monitoring – the bias and dispersion of the estimates decrease, and the blind spots in the monitored area disappear. Moreover, the fusion of data may even decrease the bias and dispersion of the estimates of other healthcare-related parameters and – what is the most important – this capacity is not affected by the temporal occlusion of the monitored person resulting in data fragmentation.
Even though the Project was oriented towards measurement-and-computer engineering, and its scope was constrained to the synergistic use of two non-invasive monitoring techniques – viz. techniques based on radar sensors and depth sensors – such projects, in general, involve interdisciplinary research carried out at the interface of technology, health-related disciplines and social sciences.
The development of accurate and fast algorithms for data fusion will directly influence further development of the techniques of unobtrusive monitoring. When the usability of the radar-sensors-based and depth-sensor-based techniques is proven and well-grounded, further advancements can be introduced, and it has to be stressed that these techniques have much more to offer: radar sensors can detect heartbeat and breathing of a person, while depth sensors can be efficiently used to detect person’s falls. Further research on the fusion of data from those sensors may lead to the development of a complete system for monitoring of elderly and disabled persons; therefore, the Project will indirectly enhance the quality of life of numerous persons.