Big data, big opportunity yet a bigger complexity
In the year 2017 there were close to 3.7 billion internet users in the world1. In the same year, it was reported that 2.5 quintillion bytes of data is generated every day2. The mobile data traffic volume worldwide is 3,000GB every second3.
In the world of the Internet of Things, everything is producing data. Depending on how detailed the data is, it can be used to create something as big as a digital twin as well. The more the number of sensors planted on a machine, the more is the data collected from it. Starting from predictive maintenance to making the machine stop, from thousands of miles of distance, everything is achievable just by looking at the data produced by the machine. The next level is making an autonomous machine, a machine that can start/stop itself, control its own functioning. Such a machine will feed on data produced by its predecessors and make decisions. Self-driving cars is one such machine that is the buzzword today.
Data produced by cars
Any car produces data and the electronic control units in a car use a lot of that in their functioning. The OBD2 port is one way of accessing this data and making maintenance and other decisions. In the context of autonomous cars, over and above this data, the car is laden with sensors like LIDARs, radars, video cameras, ultrasonic sensors etc. which capture the environmental data. The video that this camera/LIDAR records has a size of close to 36 to 72 GB per hour. This video is rich in unstructured data and, in the context of autonomous cars itself, can be consumed in various ways. Firstly, to strengthen the AI algorithms converting a car to an autonomous machine or to create these algorithms. The data is, basically, interpreted as various “scenarios” that the car encounters and keeps adding to the existing scenarios, and makes the car learn the correct response. This improves the accuracy of the autonomous decision-making system in the car. Secondly, this data is consumed in creating and updating street views and HD maps. The same video can be used to test and validate the autonomous system as well.
However, the question really is: Why is this data being used only in the driving or transportation context? We are, in essence, limiting ourselves to use the data captured by a car in the context of a car alone. It is time that we understand that that same data can be used in other contexts as well.
Contexts in which cars’ data can be used
Scenario 1 is of a car going to and from somewhere in the same stretch of a city throughout the day. This vehicle, if it is autonomous or semi-autonomous, captures the video of the same roads. Data captured this way can be used to give an idea of the demographics of the people in the area at different times of the day. Information of this kind can be fed into targeted advertisement systems such as LED/LCD hoardings in the area. For instance, an area may see school-going children at noon time, so displaying advertisements of toy brands and video games is sensible; at another time there may be more of employees of some company, probably real estate advertisements will make more sense then.
Scenario 2 is the car covering ground in a particular State, capturing details on the kind of road infrastructure. The road infrastructure will include and not be limited to traffic signal, road signs, lane marking types, freeways, the exits, merging etc. This data will in turn be used by the Government agencies to make important decisions on, say, consistency of the infrastructure.
Day to day information of this kind may be used to make decisions on the urgency of upgrading infrastructure in a particular area of the State. Technology companies and start-ups may use this same data to create better infrastructure solutions. These solutions could target problems like traffic congestion in cities or the availability of parking lots.
Data accumulated over the years: Possibilities
In a world where technology disruptors, like Waymo, Uber, NutoNomy and almost all major automotive companies, are collecting data day in and day out, it is exceedingly important to acknowledge that the data collected by them can be potentially applied across many contexts. This is also an alternate accelerated approach towards data monetization. For instance, the widely accepted industry leader Waymo (back then Google), which has been collecting data on the roads in the past decade, gets its major advantage from the data that it has accumulated over the years. Waymo cars have driven more than 8 million miles on public roads in 25 cities4. For autonomous driving systems, the more the data they learn from, the more accurate they become. However, if regulations are passed mandating the sharing of this data in the favour of public safety, then this advantage is lost. Analysing other contexts in which the data can be applied ensures that the collector of the data needn’t share the whole data but using video steganography can selectively “sell” chunks of data. This way they and other companies in a similar situation will not lose their competitive advantage, yet benefit in other ways, out of the head start they have.
1 How much data does the world generate every minute?
2 How much data do we create every day? The mind-blowing stats everyone should read
3 The global state of the internet in April 2017
4 On the Road
“The views expressed in this article are mine and my employer does not subscribe to the substance or veracity of my views.”