Right recipe for smart: add a pinch of AI to the IoT

Smart cities are already using real-time data to improve their operations. Adding intelligence and edge computing elevates the ‘smart’ concept beyond convenience and into the realms of autonomous and automated systems offering protection, saving costs and contributing to a sustainable future. A team of US universities, led by Carnegie Mellon University’s Metro21:Smart Cities Institute in Pittsburg

h, develops and deploys smart city projects in collaboration with communities and developers. The research projects have spawned real-world applications, technologies and even spin-out companies. One example is the city of Pittsburgh’s snow ploughs. Drivers were issued with iPads equipped with digital maps and audible directions to replace the paper maps they had been using. The dynamic system not only alerts drivers to road closures and suggests an optimal detour route to get to the targeted neighbourhood. It also adjusts as weather conditions change to apply more or less salt, as required. The city has reported saving $1m (£790,000) on the use of salt. Even with the cost of recalibrating the salt spreaders, Mayor Bill Peduto says: “This technology is already paying for itself and will only lead to more savings year after year.” The road data was gathered using Surtrac, a technology developed by a Metro21 spin-out company, RapidFlow Technologies. It uses artificial intelligence (AI), machine learning (ML) and traffic data from radar and cameras to enable it to adapt in real-time to changing traffic conditions. Surtrac is not confined to arterial routes and is able to optimise routes for cars and buses, cyclists and pedestrians. It has been shown to reduce travel time by up to 25 per cent, waiting time at traffic signals by 40 per cent and emissions by 20 per cent. The same traffic planning was used for what Metro21’s executive director, Karen Lightman, describes as “a smart city project, solving real-world problems”. A food delivery project in an area just outside Pittsburgh has ensured that children who qualify for free school meals have still received them despite schools being closed due to the Covid-19 pandemic. Metro21 created a machine-learning algorithm to draw up a route which included existing bus stops within a walkable distance of children’s homes. The data of   recipients was ‘scrubbed’ clean so no individual or family was identifiable, but the geo-coded data was used to optimise a route for this particularly hilly district. Drivers were able to provide feedback, such as identifying a better intersection to stop for a delivery, to further improve the service. This introduces the element of the human feedback loop to improve the route algorithm and the supply chain, as the driver is also able to confirm the number of meals that have been collected. “People have to trust the IoT, so we are thinking about the human feedback loop and building that into the IoT,” explains Lightman. “As things become more automated and eventually autonomous, if you don’t have the human feedback loop, the ‘things’ are going to optimise for the ‘things’ not for humans.” For example, if you request a service, the automated system may respond that this is ‘inconvenient’ unless you want to pay a premium rate. The human feedback loop, says Lightman, adds evaluation to provide some choice of agency. Power consumption, interconnectivity and access to broadband are vital components for an effective IoT. Rather than captured data being stored in and sent from the cloud, edge computing processes the data on a device at the edge of the network. It is a more secure, power-efficient method for conveying only part of the information. There is also an environmental factor, Lightman notes, as there is no need for supercomputers 200 miles away. “I am a little discouraged that we are not there yet,” she concedes. “I know we have the technology capability. We are demonstrating use cases of edge computing, but we are nowhere near wide adoption because we don’t have the infrastructure required. There still needs to be broadband.” The French network operator Sigfox takes a different approach. It provides a network designed to carry small data payloads (12 bytes) from IoT devices. This is sufficient to transmit temperature, humidity or GPS co-ordinates, for example. “It is a small amount of data and as broad as possible,” explains Laurent Soubielle, Sigfox’s IoT agency director. “There is no pairing, no IP address, no mobile network,” he says. The network uses part of the ISM (industrial, scientific, medical) bandwidth so it is a free spectrum worldwide. “We say we don’t need a highway to ride a bicycle,” Soubielle says. It is also resilient to noise, he adds. Power consumption is minimised because there is no network attachment. It does not have the conventional mobile device dialogue with the network as it messages its presence and then prepares to connect. Using the Sigfox network, the device switches on the RF module, sends a message and then shuts down. This short exchange saves power and ensures that the battery life is predictable. “Our goal is to connect the world,” declares Soubielle. The company provides the Sigfox1 network across 72 countries and 17 million registered devices. It has also partnered with Eutelsat to support the 0G network, using ELO (Eutelsat Low Earth Orbit for Objects) nanosatellites for coverage into remote areas, including oceans. “The IoT does not know frontiers, it is not limited to one country,” says Soubielle. The constellation of 24 nanosatellites will be dedicated to remote areas or specific use cases that are not addressed by the existing network. In Zimbabwe, the Sigfox Foundation has replaced tracking devices that were used to monitor rhinos, as they were being hacked by poachers, revealing where the animals were. Inserting a small sensor with GPS tracker in the horns of the rhinos, the Now Rhinos Speak project was able to track the animals via three signals sent each day over the secure Sigfox network. Three base stations were deployed on mountaintops across the region, with solar panels to power the network. The next step is to remove the gateway and cover the area with a satellite, continues Soubielle, so that any unusual location or movements can be spotted in time for someone to investigate if this is due to poachers. The same 0G technology is used to extend the IoT across oceans. Plastimo is a French company that makes life jackets for sailors. It will integrate a hybrid geolocation device that will be trackable tens of kilometres from coastlines where there is a Sigfox terrestrial network and more widely as the ELO constellation comes into use. The low power consumption means that an extra battery is not required, which could affect the weight or fit of the life jacket. The Covid-19 pandemic has seen a rise in interest in any devices that can avoid human contact. Dr Khasha Ghaffarzadeh, research director at IDTechEx, believes that the pandemic is accelerating the development of automation. “Anything that allows a touch-free logistics chain is of interest,” he says. The lockdown measures have seen robots being put to use across cities for food delivery services, contactless information and security and sanitation. Pavement delivery robots are regularly used in some US states and were part of a trial by Deliveroo in the UK. Starship pavement robots were on the streets of Milton Keynes. Such robots are constantly monitored and a human takes over control if it needs to cross a street or to find a place to ‘park’. They are limited to speeds of about 6km/h (3.7mph). Today, they rely on algorithms to navigate and learn local areas using cameras, which are cheaper than lidar (light detection and ranging) technology but whose performance can be affected by weather conditions. Sidewalk or pavement robots are mainly confined to known environments and deliver relatively low-value items. In time, robots may be used for goods ordered online, with the purchaser receiving a code on their phone to unlock the robot’s container when it arrives.   Ghaffarzadeh believes that, once the blind spots which can blight lidar systems are ironed out, the industry will need to refine the design of the container (to carry more cargo), battery capacity and range. He is optimistic and believes “in five to 10 years’ time, sidewalk robots will be fully autonomous”. Robots are also being repurposed. The Chinese agricultural robotics company XAG has deployed its R80 crop-spraying robots to disinfect public areas. The robot minimises human contact to prevent the spread of disease and can disinfect a wide area, systematically via its jet spray nozzles, which rotate 360° for complete coverage of 3D objects. Its pre-programmed route covers areas where lorries cannot fit, using sensors to navigate.   The IoT may be a mature technology but the addition of edge computing, AI and ML will enhance a connected environment, creating one that engages humans while improving quality of life.

                             Image credit: Shipping Technology
        
  


  Marine
  From the roads to the waterways


  Autonomous canal boats using AI and data gathered by the vessel’s onboard systems are being developed by Shipping Technology. The Amsterdam-​based automation company specialises in marine applications and is working with AI consultancy Xomnia to develop and implement AI using data from the boat’s motor management, radar and GPS. The data is gathered and processed using the Black Box Pro computer and software installed onboard.

Black Box Pro connects to all nautical systems and gathers data from them which is stored in the cloud. It also runs applications such as a collision-detection system and autopilot functionality. The system is designed to monitor and control ships and support safe and reliable autonomous shipping. The 10m-long test boat was able to send data to ship owners, providing real-time updates of the cargo as well as historical data via a dashboard view. The semi-autonomous vessels are able to develop algorithms based on the data gathered via navigation and systems management, together with the captain’s behaviour and reactions in situations. The project aims to build a system that has learned from hours of sailing by experienced captains on the region’s inland waterways. There are also sustainability benefits, as the data can be analysed to plan the most effective routes and navigate to optimise fuel consumption depending on the load. Managing the data from crews, ports and operators can also be used to streamline loading and unloading and for predictive maintenance to maximise sailing time. There are 50 vessels equipped with Black Box Pro, which has been beta-tested and is now ready for early adopters. Shipping Technology reports that its vessels have sailed over 150km fully autonomously.

  Safety
  Going underground


  A little-publicised danger faced by city dwellers is exploding manhole covers due to bad weather or flooding damaging underground power systems. Between 2010 and 2015, the UK Health and Safety Executive reported 80 incidents of manhole explosions as a result of faults in underground power lines. There are many more reported in the US.

UK engineering and design company Plextek and US-based CNIguard have collaborated to develop the Sentir system, which has been deployed in New York by energy supplier Con Edison. A sealed device is attached to the underside of the access cover using a magnet or a customised bracket. Inside is an accelerometer, which detects movement, and sensors to detect flooding and monitor gas, stray voltages, arcing, temperature and humidity. The unit transmits signals to a dashboard displayed on a laptop or mobile device to warn of situations that need attention and so prevent problems.The system can also be used to secure public municipality or utility underground boxes, chambers, vaults, cabinets, kiosks and bunkers to protect them from theft or vandalism. Water pipelines, too, can wreak havoc when damaged. Last summer, Vodafone and South East Water launched a pilot scheme to monitor underground mains water pipes for leaks, and in June 2020 signed a 10-year agreement with SES Water for a similar system. A network of water meters, sensors and acoustic loggers collect data on the pipework, listening for any sounds that indicate a water leak. Data is transmitted via the phone operator’s NB-IoT narrowband network, giving a precise location of the leak.

Author: Caroline Hayes

Date: 2020-11-11

URL: https://eandt.theiet.org/content/articles/2020/11/right-recipe-for-smart-add-a-pinch-of-ai-to-the-iot/

theiet.org