Article courtesy of Innodisk
In a previous article we defined what the terms IoT, AI, AIoT and Edge Computing mean. Now let us look at some examples of how these would work out in real-life scenarios,
City Traffic Surveillance
Our cities are growing in three dimensions by spreading outward and upward (by buildings growing in height). Roads, however, are still mostly confined to two dimensions, which leads to increased traffic congestion as cities grow larger.
Monitoring and altering traffic flow based on real-time data can significantly increase efficiency and cutdown congestion. This an be done with surveillance installations strategically placed throughout the city.
The first-step analysis is handled by local AI platforms at the edge. This includes vehicle recognition and traffic flow assessment. Each installation can thus determine by itself how to handle the data based on the analysis; i.e., is the number of vehicles increasing and is there a risk of congestion? Any essential data can then be sent to a centralized platform (the cloud), where measures such as redirecting traffic, altering speed limits, and adjusting traffic lights can be taken based on the data.
Fleet management and AI
AI can significantly optimize fleet management operations. Monitoring a large fleet of vehicles can be hard but there are many ways to streamline operations: reducing fuel costs, vehicle maintenance, mitigating unsafe driver behavior etc.
The current positioning systems are mostly reliant on GPS, which fails to handle certain problems. For example, entering a tunnel renders the GPS all but useless and the system will have no idea where the vehicle is located. This also happens within cities when driving inside buildings or other areas with poor satellite coverage. It is also difficult for the system to determine the vehicle’s elevation.
However, there are other sources of data other that can give us a pointer on vehicle position: a vehicle’s speed and turning rate can be constantly monitored and logged. An on-board AI platform can then calculate the vehicle position is at any point in time by having these parameters compensate for incomplete GPS data. This technology is called automotive dead reckoning, or DR. Lastly, data can be transmitted through wireless networks back to the operator.
Autonomous Delivery Robots
When we remove the human factor from vehicles, the main problem we run into is the ever-changing traffic picture that is fraught with unexpected factors. Because of this, an autonomous vehicle has to be able to make split-second decisions with any sudden change happening in its path. Where we rely on our senses, the robot has a multitude of sensors that gather all kinds of data that has to be processed into a coherent image of the overall situation at any moment in time. Relying on the cloud, in this case, is hopeless as the latency will surely mean that by the time the data is ready and a decision can be made it is already too late.
The on-board AI platform that handles these complex calculations is reliant on components that work under whatever weather and physical conditions are present without any drop in performance. To avoid accidents involving autonomous vehicles it is prudent that the equipment is performing with minimal chance of failure and with sufficient backup.