How Artificial Intelligence technology is used by traffic apps
Long gone are the days where we had to use physical maps to navigate between destinations. Nowadays, everything can be done with the touch of a button on your smartphone. And, although it may seem like magic, your maps and traffic applications can now predict what the best route to take is, where trouble spots are building up, alternate routes and your estimated time of arrival. How is this possible? It’s all thanks to Artificial Intelligence and this blog looks at the role AI and Machine Learning play in perfecting your route.
What is Artificial Intelligence and Machine Learning?
Artificial intelligence makes it possible for machines to learn from ‘experience’, adjust to new inputs and perform tasks. Whereas, Machine Learning, is the software and algorithms that make machines smart, or the science of getting computers to act without being explicitly programmed.
For predicting routes and traffic, algorithms are used to decide how fast your journey will be on the routes it suggests. This is based on factors like paved roads, motorways, and time of day – all historical data that will be taken into account when you type in your destination. But while this information helps you find current traffic estimates —whether or not a traffic jam will affect your drive right now—it doesn’t account for what traffic will look like 10, 20, or even 50 minutes into your journey.
This is where artificial intelligence helps. By looking at live traffic conditions AI can help your Maps application adjust in real time and predict whether traffic will likely become heavier in one direction as a result – if so, the app will automatically find you a lower-traffic alternative.
What data is used?
To predict your arrival time and the best route you should take, a wealth of data needs to be analysed from various sources and fed it into machine learning models to predict traffic flows. The sort of data that is used includes live traffic information collected anonymously from smartphones, historical traffic data, information like speed limits and construction sites from local governments, and also factors like the quality, size, and direction of any given road. What’s more, with apps like Waze, users can also contribute real-time data to the Waze Community, for example pointing out where the police are checking driving speeds and if a traffic accident has occurred.
However, it is worth noting that take part in this data collection is entirely optional. Only those who are willing to share their information with the app will be contributing. If you’d prefer, you can disable the Location Services on your phone and none of your data will be accumulated.
To predict what traffic would look like in the near future, Google Maps and other similar applications, analyse historical traffic patterns for roads over time, using data from past trips down this route. For example, one pattern may show that the 280 freeway in Northern California typically has vehicles traveling at a speed of 65mph between 6-7am, but only at 15-20mph in the late afternoon. The company then combines this database of historical traffic patterns with real time traffic conditions, using machine learning to generate predictions based on both sets of data. Using AI, the company says they can accurately predict your estimated time of arrival for over 97% of trips. What’s more, the more you use your preferred app the better it gets to know you. By learning your frequently used routes and destinations, as well as the hours when you commute, your fastest route can be further optimised.
With many people working from home and going out less often because of the coronavirus, companies have also updated their AI model to prioritise traffic patterns from the last two-to-four weeks and deprioritise patterns from any time before that. As a result, Maps will already know that a route that might have taken 30 minutes in rush hour, now may only take 20 minutes, and will be able to update the estimated time of arrival accordingly as more people start driving again.
Predicting traffic and determining routes is incredibly complex. However, as more data is added to the set used to make predictions, the accuracy rate is only likely to improve. You can read a detailed blog from Google on the subject here.
Interested in other technology involved in travel? Why not read one of our other blogs: