Ask any soccer fan and he or she will tell you that 2014 World Cup was one of the best in many years. We can list many reasons for that. Together with the 1998 edition, this World Cup averaged the highest number of goals since 1970. Exciting matches were decided in the last minutes with some surprising results, and teams that don’t traditionally do well in international soccer competitions did great, like Costa Rica, which at the end of group phase was leading a group with three world champions. Nonetheless, one specific event will be forever remembered from the 2014 World Cup edition in Brazil. And no, I’m not talking about the huge 7×1 win from Germany against Brazil in the semifinals. I’m talking about the usage of technology to support referee decisions for the first time in World Cup history, more specifically the so-called goal line technology.
The day was June 15. France and Honduras were debuting in the competition. France was already winning the game by 1×0 when, in the second minute of the second half, French striker Karim Benzema shot a ball that hit the post and continued parallel to the goal line. The ball was intercepted by the goalkeeper, whose position caused doubt among viewers as to whether the ball entered the goal or not. At the same moment, the match referee validated the goal without hesitation, thanks to the smart watch he has wearing. The device confirmed the goal was valid and a few seconds later the image replay confirmed the ball crossed the line completely, characterizing a goal.
For American football or tennis fans this would be just another instance in which a game was decided with technology support, but for soccer this is a brand new concept. Goal line technology was piloted for the first time in 2013, during the 2013-14 season of the Premier League and during the Confederations Cup in Brazil. Once the first results were validated, FIFA approved the technology to be used for the first time in the most important soccer competition, the FIFA World Cup.
So how does this goal line technology work? Its official name is GoalControl-4D and the technology is based on a research area named computer vision. The general idea behind it is to detect and keep track of moving objects in a tridimensional scene that is rendered using images captured by video cameras. In a soccer field, the tracked object is obviously the ball and the scene is bounded by the lines used to define the field area.
The technology used by FIFA relies on 14 high-speed cameras capable of capturing 500 frames per second (a regular camera captures about 30 frames in a second). Cameras are positioned around the goals (seven on each side) at the stadium roof. A software algorithm processes the images generated by the cameras, rendering the scene and calculating the ball position in three dimensions every two milliseconds. This allows the system to know the ball location with a precision of millimeters, even when the ball is not visible for all 7 cameras. The high precision location system allows the software to know exactly if the ball crossed the goal line or not. In addition, when a goal is detected the system sends a notification to the smart watch used by the referee to confirm the goal.
If this kind of technology is transforming professional sports, you can imagine its potential when applied to the industry to improve efficiency and increase productivity.
Let’s take the mining sector as an example. Brazil is currently the third biggest iron ore producer in the world, with more the 400 MM tons per year. Most of this production is exported, representing 15% of all the exported goods from the country. When considering the process of iron ore extraction, one of the most important stages is the transportation logistics, either inside the mining site or outside, when moving the ore from pit to port to be exported.
One of the common solutions for ore transportation inside the mine is conveyor belts. In certain locations you can find kilometers of lines used to transport the ore between the different stages of production and the storage areas (see image below). Conveyor belts can move tons of ore every day, being the mine equivalent to main avenues in big cities like São Paulo. This means that if any of these “avenues” is interrupted by an obstruction or a rip on the belt surface the potential losses might be considerable, with chances of causing a plant shutdown in some extreme situations. Such incidents may occur because of natural belt surface wear out or be caused by unpredictable situations like a foreign body entering into the conveyor circuit. Foreign bodies can be broken parts of tools used in the stages of the process, pieces of the guardrails installed around the conveyor belts or even ore rocks with bigger sizes or sharp edges that passed through the sieves.
Automatic detection of such incidents in conveyor belts is a problem that many mining companies and solution suppliers are trying to solve using different techniques. A common goal is to be able to identify the problem in the early stages of the process, reducing impacts in the production and the risk of a plant shutdown.
So what about using the same cameras that monitor the ball in the soccer field to identify if there are foreign bodies in the conveyor belt? Or even predict rips that are about to happen by analyzing the texture of the belt surface or its temperature?
Once the expected granulometry is known for each stage of the ore production, this information can be used to define the patterns used by the image analysis software. Based on that, the system could send a notification to the nearest referee (or field engineer), according to the location of the conveyor belt, when something unexpected happens. To detect the wear out of the belt, thermal cameras also could be used to analyze and identify temperature variations caused by friction with fixed parts of the system that might result in belt rips.
There are of course challenges when applying such technologies to industrial usage, especially in harsh environments like mining pits with high concentrations of dust. Such environments can compromise the camera detection capabilities or require a rigorous maintenance routine to clean up the accumulated dust in the camera lenses in a long term. Solutions like self-cleaning cameras or even more robust technologies based on infrared detection or lasers also can be considered despite having higher costs. With sensors prices going down so fast it is just a matter of time for such solutions to become popular in more markets.
This is only one example of the potential uses of computer vision technologies in the mining segment. There are many others including equipment integrity check, speed or distance measurement and even reservoir volume monitoring. All the examples mentioned can be definitely ported to other industry segments, specially the Oil & Gas industry that is also very important for the Brazilian economy.