Identifying suitable Hardware is often a main hindrance factor for the successful implementation and scaling of an IoT Use Case.
Smart Waste is one of the key Use Cases in the Smart City ecosystem and widely discussed and piloted by various cities and entities. Technically though, the case is not as simple as it sounds and as might be expected. Varying sizes and forms of waste bins and heterogeneous contents lead to falsified data and unclear results, factually disabling scalable solutions. Most “off the shelf” devices currently work based on ultrasonic-sound technology. Rebouncing ultrasonic interferences in the signal due to different bin-designs or its content can hardly be diagnosed.
The team of Dr. Luciano Sarperi and Patrick Rennhard of the Zurich University of Applied Sciences in Winterthur have taken a different technical approach by using upcoming low cost radar devices. This publication of their research will hopefully be adapted by sensor manufacturers and improve the quality of smart waste devices in the future, finally enabling high quality results and scalable smart waste solutions.
Smart Waste Management with Radar Sensors
The Institute of Signal Processing and Wireless Communications of the ZHAW School of Engineering has studied the feasibility of using low-cost short-range radar sensors for reliable fill level measurement in smart waste-management applications as an alternative to ultrasonic sensors.
Radar sensors can potentially provide a reliable measurement of the waste fill level since radio waves penetrate non-metallic waste and can therefore measure the presence of waste from the waste surface down to the bottom of the bin. This can be helpful to determine the fill level for low-density waste, which does not compactly fill the waste bin.
Low-cost short-range radar sensors, starting at just a few euros, with antennas-on-package operating in the 60 GHz band are now becoming available. They are relatively low power and can be classified as either Frequency Modulated Continuous Wave (FMCW) radars, which allow a minimum measurement distance of virtually zero, or pulse radars, with a minimum measurement distance of a few cm.
We carried out initial studies with a metallic waste bin with 70 cm height and 45 cm maximum diameter using a Texas Instruments AWR1642 automotive radar in the 70 GHz band set to an opening angle of 30⁰. First tests with an empty bin showed that the radar could reliably detect the distance to the bottom of the bin, without undue interference from reflections off the walls.
Next, tests with mixed waste were conducted to determine the fill level. Figure 1 shows an example with the bin at 80% fill level and having an irregular waste surface.
The resulting power vs. range profile in Figure 2 shows that the peak powers match well the real distances from the sensor to the top level of the waste (standing cardboard box) and to the mean fill level. Additionally, the presence of waste below the mean fill level can be detected, down to the bottom of the metallic bin. The bottom of the bin can be clearly distinguished due to the large amount of reflected power.
To implement a reliable automatic fill level measurement system, the raw power vs. range data should be further processed with an algorithm to make use of the complete available information, instead of only estimating the distance to the nearest or largest object.
For practical implementations, the opening angles of radar sensors with antennas-on-package can be too large. In such cases, a low cost solution to reduce the beam width is using an optical lens. We have successfully produced such lenses for the 60 GHz band using a 3D printer with polylactic acid as a printing material.
With low cost short-range radar sensors becoming available, new possibilities arise for implementing cost effective and reliable waste fill level measurement systems.
About this project
This work was carried out in the context of a bachelor student project for the ZHAW School of Engineering and has been supported by Akenza AG.
Publication of Sarperi, Luciano; and Rennhard, Patrick; Hermann, Adrian, 2019: Narrowband IoT in der Praxis.