Addressing the challenges of accurate people counting using mmWave radar

Automated people counting systems, which measure the number of people passing through an entrance, first appeared in the 1980s and their use has grown steadily over the years. The retail sector has traditionally been the biggest user of these systems, deploying sophisticated retail analytics systems in their efforts to better understand their customers.

The use of people counting systems outside of retail has seen an acceleration as an increasing range of businesses in sectors as diverse as transport, banking and financial, hospitality, healthcare and sports and entertainment strive to increase efficiencies through the use of data.

People counting systems generate data which give valuable insights into business operations. Retailers measure visitor count and behavior at their stores and use the data for functions such as staff planning and to measure the effectiveness of marketing campaigns. Local government can use people counting systems to reduce energy costs in public buildings such as schools and libraries, adjusting temperatures and lighting levels to match building occupancy. Effective use of space is a priority across all sectors and people counting systems can help identify underutilized areas, contributing to rental cost reduction.

Already experiencing continuous levels of growth, the demand for people counting systems surged in 2020 because of the COVID-19 pandemic. Social distancing requires careful management of occupancy levels in public spaces, offices and retail outlets – and people counting systems enable the smart management of these levels so that facilities can operate safely.

Existing automated people counting technologies

Modern people counting systems use some form of electronic device to detect when a person crosses an invisible count line. Most solutions nowadays are equipped with interfaces which enable their collected count data to be transmitted to back end systems, such as retail analytics platforms, over the internet or any other communications network.

The shrinking cost and form factors of integrated electronic devices has led to the emergence of a wide spectrum of sophisticated, low-cost sensors, which generate electronic signals in response to a variety of physical stimuli. As the demand for more accurate data has grown, designers of people counting systems have found increasingly innovative ways of leveraging these devices to improve the reliability and accuracy of their solutions. The table below gives an overview of the most common systems in use today, each offering different levels of accuracy at various price points, corresponding to the specific needs of different counting applications.

The first automated systems to appear on the market transmitted infrared (IR) beams across the entrance, which were broken by passing people or objects. Automated people counting systems based on this technology offer around 85% accuracy rate and are still widely used. However, these systems require careful mounting, performance degrades in direct sunlight, and do not offer the same level of accuracy with crowds of people.

Compare and contrast of people counting technologies table

 

Image-based counting systems are the most popular of the current generation, taking advantage of the proliferation of small, low-cost but high-resolution video cameras on the market. These systems offer counting accuracies of up to 98% and have other advantages, such as their ability to track individual customers as they move through the store, generating rich information for retail analytics applications. Image-based systems are however relatively complex and costly to install. It also requires sophisticated image processing software, leading to a high total cost of ownership (TCO). Video systems may also encounter data protection and privacy issues depending on local regulation.

Thermal imaging sensors, which detect heat sources from the human body, are also commonly deployed in people counting solutions. These solutions can provide accuracy levels up to 95%, even with high traffic volumes, but require steady traffic flows. People stopping in the counting area can lead to false counts, reducing the accuracy rate.

Modern people counting systems also employ other technologies such as Bluetooth. As counting is triggered by the number of mobile phones in the detected region, the system provides a relatively low accuracy rate and only those with mobile phones will be counted.

The next generation of people counters

A new generation of sensors have emerged in recent years, based on developments in the automobile and industrial sectors. Radar and time-of-flight (ToF) are used to detect and analyze objects within their field of view. ToF imaging sensors use VCSEL to emit laser beams, which are reflected back to the sensor by any object in its path, while radar sensors use radio waves to detect objects. The narrow beams from lasers give high resolutions and accuracies at short distances, but resolutions diminish over longer distances as the beams widen. ToF sensor accuracy is also adversely affected by bright reflections and certain surface types. Radar sensors use radio waves rather than lasers to detect objects and can operate accurately over longer distances than ToF sensors. They are also less susceptible to poor weather conditions and reflections.

Radar sensors operating in the mmWave spectrum are particularly effective at detecting people and objects through materials such as glass and drywall, enabling them to be easily concealed. These sensors are highly effective at detecting small movements in poor environmental conditions, such as rain, smoke, rain and low light. Readily available on the market in compact form factors at relatively low cost, mmWave radar sensors bring exciting new possibilities to the field of people counting.

Choosing the optimum technology for a particular people counting application requires careful consideration of factors such as the requirements of the application itself, the environment in which it will operate and its communication and interfacing requirements. TCO must also be considered, which will be based on the capabilities of the enterprise and the required support environment.

People counting challenges

People counting systems are required to operate in a number of different scenarios as laid out in the table below. These scenarios, along with the required counting accuracy, determine which technology is optimal.

Compare and contrast of people counting scenarios table

At the simplest end of the spectrum, the traffic flow through an entrance is managed such that only one person at a time passes through and there is a minimum distance of 1.5 meter between each person. Most technologies can cope with this scenario but, in reality, most entrances cannot be so rigorously controlled. People may pass in either direction, in groups or close behind one another. Pushchairs and shopping trolleys introduce further potential for confusion.

Most technologies can deal with scenarios 1 and 2 but simpler systems, such as IR break-beam and Bluetooth, will lose accuracy in more complex situations, such as scenarios 4 and 5. Although thermal imaging and video systems work well across all scenarios, their high installation and operating costs may be prohibitive, particularly where a highly cost-effective and accurate solution is required for rapid deployment. The benefits of mmWave radar sensors address some of the challenges: they are relatively cheap, easy to install and their small form factors enable discrete mounting, avoiding potential customer concerns over surveillance.

A new smart entrance counter solution

Leveraging its XENSIV™ 60 GHz radar sensor, Infineon recently launched the Smart Entrance Counter Solution that address the need for a size-, cost- and performance-optimized solution.

Infineon 60Hz mmWave radar sensor module
Infineon's Smart Entrance Counter Solution leverages the XENSIV 60 GHz radar sensor.

The solution’s design emphasizes flexibility and ease of use and is a true turnkey solution, coming packaged with fully tested and qualified software. A wide choice of connectivity interfaces, including Wi-Fi, Bluetooth, Bluetooth Mesh, UART, USB, SPI/I2C and GPIO, provide flexible options for integration with anything from local traffic lights to back-end cloud systems (Figure 3). Solution access is enabled via an intuitive Graphical User Interface (GUI). System testing and integration is further simplified by the module’s FCC certifiable reference design, which means that installation and testing does not need specialized RF resource.

People counting solution using mmwave radar block diagram Infineon
The Infineon People Counting Solution integrates easily with customer platforms.

 

Available in a variety of mounting options, the bidirectional sensor comes ready-configured to meet the challenges of the first three scenarios outlined above. Accuracies for these are in excess of 98%, and Infineon plan to release future versions which will address more complex scenarios.

Already growing in popularity, the importance of people counting systems, has perhaps never been higher than in 2020/21 due to the COVID-19 pandemic. Thanks to an evolution of technology, users have a wide range of options on the market and must choose carefully based on the budget, required accuracy and the environment in which the system must operate.

Additional Background

Infineon’s Smart Entrance Counter Solution, which enables accurate, anonymous and contactless people counting, was selected as a CES® 2021 Innovation Awards Honoree in the "Smart Cities" category. The CES Innovation Awards program, owned and produced by the Consumer Technology Association (CTA)®, is an annual competition honoring outstanding design and engineering in consumer technology products across 28 product categories. An elite panel of industry expert judges, including members of the media, designers, engineers and more, reviewed submissions based on innovation, engineering and functionality, aesthetic and design. CES Innovation Awards are granted to products based upon materials submitted by the honorees. CTA neither verifies the accuracy of any submissions nor tests any products.

 

RELATED

mmWave-based radar expands use cases for people, motion sensing

Scope allows mmWave measurements to 110 GHZ