Patrick Groth

Advisors
Nils Roth (M.Sc.), Prof. Dr. Björn Eskofier

Duration
1/2020 – 06/2020

Abstract
Smart homes play a major part in today’s digitalization and are enjoying increasing popularity over the last years [1]. The basis of every smart home system is motion detection, which is used to lower power consumption, make living more comfortable and increase the safety of homes by controlling the heating, ventilation, air conditioning or lighting system as well as to detect intrusion. The main problem of today’s motion detection is that they are either too inaccurate or too accurate and thereby either register movement without sufficient reason or not at all. Also, they don’t take other factors into account like the direction of the movement or the type of object. Another problem is that they can’t recognize stationary humans or ignore irrelevant moving objects like fans, leaves or pets [2]. The reason for this is that they mostly rely on passive infrared sensors to detect movement [3]. Due to their simple design and low cost, they lack accuracy as well as any type of distinction. Other solutions, like ultrasonic or microwave sensors, pay attention to special sounds or the reflection of waves to check if objects are moving or not. While those sensors are far more accurate in detecting movement, they also lack any type of interpretation and thereby trigger wrong easily. Latest sensors based on time of flight and radar technology promise, when combined with classification approaches, a solution to these problems [4]. For instance, pets should not be able to trigger a motion detector, whereas small children and other humans, even when sitting relaxed, should trigger the sensor. Furthermore, a person’s direction or intention should be detected, to e.g. distinguish between a person going up or down a staircase. As shown in [4] and [5], the classification of humans via radar sensors with little power consumption is possible. Kiuru et al. [6] also presented an approach to detect respiration and intrusion via radar sensors. While there has been tremendous progress in the area of pattern recognition and classification, as well as the development of smaller, more effective sensors, there are still multiple challenges, which need to be overcome, especially in smart-home scenarios. First and most importantly, the setup of sensors must be relatively easy and flexible, so that they can be mounted in different positions and locations, for instance, a staircase, the entrance of a house or in a room. Power consumption also plays an increasing role [7]. That’s why accurate real-time decision-making with only limited computing power is essential. Lastly, it is not feasible for users to train their sensors in order to get good results and therefore data collection is also a topic not to be ignored [8].
The goal of this bachelor thesis is to develop an algorithmic pipeline based on an available motion sensor, which is capable of classifying patterns to differentiate between different types of movements, objects and states. Based on three starter kits, the functionality of different sensor modalities shall be evaluated. Subsequently, classification approaches shall be implemented and tested. Major steps are the selection of a suitable sensor, the development as well as the selection of classification approaches and the validation of the outcome by performing a self-designed study. The system used shall be compatible with KNX, an industry standard for smart-home devices [9]. It provides the necessary standardization to make sure the sensor can be integrated into state of the art smart-home systems.

References:
[1] B. L. Risteska Stojkoska and K. V Trivodaliev, “A review of Internet of Things for smart home: Challenges and solutions,” J. Clean. Prod., 2016, doi: 10.1016/j.jclepro.2016.10.006.
[2] T. Labeodan, W. Zeiler, G. Boxem, and Y. Zhao, “Occupancy measurement in commercial office buildings for demand-driven control applications – A survey and detection system evaluation,” Energy and Buildings, vol. 93. Elsevier Ltd, pp. 303–314, 15-Apr-2015, doi: 10.1016/j.enbuild.2015.02.028.
3
[4] S. A. Shah and F. Fioranelli, “RF Sensing Technologies for Assisted Daily Living in Healthcare: A Comprehensive Review,” IEEE Aerosp. Electron. Syst. Mag., vol. 34, no. 11, pp. 26–44, Nov. 2019, doi: 10.1109/MAES.2019.2933971.
[5] E. Yavari, C. Song, V. Lubecke, and O. Boric-Lubecke, “System-on-Chip based Doppler radar occupancy sensor,” in Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS, 2011, pp. 1913–1916, doi: 10.1109/IEMBS.2011.6090541.
[6] T. Kiuru et al., “Movement and respiration detection using statistical properties of the FMCW radar signal,” in 2016 Global Symposium on Millimeter Waves, GSMM 2016 and ESA Workshop on Millimetre-Wave Technology and Applications, 2016, doi: 10.1109/GSMM.2016.7500331.
[7] F. Corno and F. Razzak, “Intelligent energy optimization for user intelligible goals in smart home environments,” IEEE Trans. Smart Grid, vol. 3, no. 4, pp. 2128–2135, 2012, doi: 10.1109/TSG.2012.2214407.
[8] S. Z. Gurbuz and M. G. Amin, “Radar-based human-motion recognition with deep learning: Promising applications for indoor monitoring,” IEEE Signal Process. Mag., vol. 36, no. 4, pp. 16–28, Jul. 2019, doi: 10.1109/MSP.2018.2890128. 9