Due to the rise of ICT (Information and Communication Technologies) and the Internet of Things, cities and companies now have the possibilities to evolve, through new techniques, better bandwidth and cheaper components. Thomas will guide you trough the Aptus journey of AI.
n the context of Smart City/Industry, we are working to develop innovative solutions to improve the operational efficiency of our customers, both in the public and in the private sector, here at Aptus. This translates in reducing the production cost for companies, and increasing the citizen welfare for local governments.
These innovations allow public and private players to harvest an enormous variety of data, in near real-time, and to store and process it within the Cloud.
In parallel to the development of new technologies for the transport and storing of the data, there has been a growth of methods to process this huge amount of data, with the aim to reduce costs and to increase the efficiency and the reliability. These methods encompass Automation, Statistics, but also Artificial Intelligence through Machine Learning and Deep Learning algorithms.
At Aptus, we have started developing solutions based on Artificial Intelligence to answer some of our customers needs. For example, we use AI with computer vision to obtain useful metrics, like the attendance on certain locations, information on the traffic or the presence of anomalies.
These solutions are great as they make the harvesting of these metrics being autonomous (no operators needed) and being totally continuous. Tough there is an operator on the far end, who receives alerts or the obtained metrics, his tasks are not the same as if he had to monitor the video streams himself.
As it is the case for other solutions within the Internet of Things, there is a choice to be made between Cloud Computing and Edge Computing for each of our solutions based on AI. So, do we choose to send the data from the sensor to the Cloud to be processed ? Or, do we choose to process the data locally, on the edge of the sensor, and to send the results to the Cloud ?
At Aptus, we have chosen to use Edge Computing. It has some benefits, as it reduces the bandwidth needed for the communication to the Cloud, and the overall latency. This also decreases the processing and storage capacity needed in the Cloud, and by such its cost and its energy footprint.
Focusing on Edge Computing also impacts security, and can make the solutions more GDPR compliant. Processing the data locally avoids the use of a gateway to send the data to the Cloud, as only the results (metrics) are sent. However, it necessitates the establishment of a secured architecture, so that no one can steal the data locally.
Obviously, the use of AI requires specific processing capacity which are not available in most common Edge Computing devices. In order to get the processing done in the Edge, we need to use hardware dedicated to AI.
Luckily, there is a variety of single-board computers which have dedicated AI processors or graphic card, with some of the most famous (from left to right): the Nvidia Jetson Nano, the Google Coral Edge TPU and the Intel NCS 2 (based on Movidius Myriad)
At Aptus, we are using the Jetson Nano and the Jetson Xavier NX from Nvidia for computer vision solutions. With their dedicated graphic card, we are able to process up to several cameras video streams, and to do Object Detection with a frame-rate reaching 30 fps in some applications. As such, with our partner, the city of Knokke-Heist, we delivered a solution based on the Jetson Nano to monitor the affluence on the beach. It processes in near-real time the video streams and send the attendance to the Cloud, where the city officials can visualize the data.
This article shows you an example of the use of AI for Smart Cities by Aptus in the domain of computer vision, particularly for surveillance. It is a fact that Artificial Intelligence can be used for numerous applications in computer vision like monitoring parkings, monitoring the traffic, autonomous vehicles, people surveillance or location surveillance (like our example).
But AI is not only reduced to computer vision. It can also be used for many Smart City applications like predicting the cost and maintenance, communicating with the citizen via bots, anonymising the harvested data, detecting anomalies and so on.
If you are interested to discuss more about Machine Learning and Artificial Intelligence, and on how Aptus can leverage these technologies for you, feel free to contact us.
