Building Smart Cities
As cities grow due to the increases in population across the world and the increased movement towards cities, problems with efficiency start to become more apparent as the systems in place aren’t ready for the influx of people. Solutions to these problems are sometimes hard to spot as they are very marginal improvements. This is the case with machine learning but these “mini solutions” can be found and applied, making a big difference with many marginal gains, in real time.
In relation to smart cities, machine learning can be defined as “systems that can sense their environment, think, learn and act in response in response to what they sense”. This article explores three ways that cities are becoming smarter with the help of machine learning.
Anyone who lives in a city will understand the issues caused by traffic congestion. It’s not just pollution and road rage. Congestion makes it more difficult for emergency services to respond to calls in an appropriate amount of time. Every small step to improving congestion in our cities is contributing to a variety of factors that could save lives.
As more of us are using public transport to travel around cities, buses are contributing to congestion problems. Video and tracking capabilities can both be linked together with machine learning to help paint a bigger picture of bus congestion so a solution can be more easily found instead of looking at the problems one by one. Traffic lights controlled by machine learning can use real-time data to change the timing in traffic lights to adjust the flow of traffic. In addition to this, machine learning can be applied to bus routes, so based on historic passenger and traffic information, buses can be diverted to avoid areas where congestion is high but passenger numbers are low.
Energy & Environment
Changes to the environment are having an effect on cities and their citizens more than ever. Issues range from air quality and pollution to flooding and fires. As a result, urban planners are faced with a new set of challenges which need to be addressed as a priority to ensure our cities continue to be safe places to live.
Machine learning has been hailed as a game changer for climate change and the environment. Although still a relatively new technology, the use of machine learning is increasingly being used to support scientists and environmentalists. IBM’s Green Horizon Project is using ML to forecast air pollution, track the sources of pollution and then produce strategies to deal with it.
It’s not just responses to environmental disasters that machine learning can assist with though; ML is increasingly being used as a means of ensuring cities are running efficiently, not using more energy than is absolutely necessary. For example, smart energy grids are being implemented which use real-time data collection and analysis machine learning to determine areas that need the most energy and areas which need energy conservation.
Applying automated video analysis to footage from CCTV systems does raise a number of ethical questions but there is no doubt that it helps sift through footage to assist police in criminal cases. The police already use license plate technology to near instantly connect a license plate to any data that they have on it. Machine learning can be applied to find patterns in criminal activity or simply to help sentencing and parole recommendations.
The term predictive policing is often used to explore the benefits of machine learning applied to policing. It might sound a bit Orwellian but it’s much less sinister than that. For example, police having access to data on the last 100 times a crime was committed in a city and when, and then being able to send police patrols to those areas around those times in case another crime occurs, has the potential to create a safer environment and save lives.
All of these systems can be linked together to provide an interconnected web of services that are constantly improved upon due to the suggestions and efficiency of machine learning.
For example, it’s possible to link together video analysis with traffic analysis to give even more detailed picture of what’s happening on our roads. For a practical example, if there are a large number of buses during certain time periods, but nobody is getting on or off these buses at certain stops, video analysis could track this. Paired with traffic analysis, machine learning could pick all these nuisances up and suggest a more efficient route to reallocate the buses to a route that requires more buses due to overfilling, in turn reducing traffic congestion.
There is no one magic solution to improve our cities and urban planning but, as this article explored, machine learning can provide gradual improvements. Over the next few years machine learning systems will be implemented into everyday city life whether you notice it or not. Machine learning has the potential to increase efficiency in our cities which could mean lower carbon emissions, lower crime rates, less congestion and the simplification of public transportation and parking.