1
\$\begingroup\$

Is it possible to use some sort of sensors (or if anything else is better please tell) to obtain a football's position within the goal posts?

Something in the order of this

What I am trying to achieve is the position of the ball as a goal is scored. Then using that position (x,y) to plot.

I have little no experience with this sort of thing (software engineer) so just wondering if its possible. If so, I would love elaboration or research links.

\$\endgroup\$
6
  • 3
    \$\begingroup\$ A couple of cameras should do it. \$\endgroup\$ Commented Feb 3, 2018 at 14:08
  • \$\begingroup\$ A potential problem is those cameras being covered by a person (goalkeeper). \$\endgroup\$
    – Sam
    Commented Feb 3, 2018 at 14:10
  • \$\begingroup\$ Are cameras the most efficient given that a ball can travel > 60mph? \$\endgroup\$
    – Sam
    Commented Feb 3, 2018 at 14:49
  • 1
    \$\begingroup\$ Step back and explain what problem you are really trying to solve. Do you really need that position of the ball within the goal, or are you just trying to determine whether the ball is in the goal such that a goal was scored? What's the overall purpose? \$\endgroup\$ Commented Feb 3, 2018 at 14:50
  • \$\begingroup\$ Thanks @OlinLathrop, I've added an extra sentence which hopefully explains. \$\endgroup\$
    – Sam
    Commented Feb 3, 2018 at 15:00

3 Answers 3

1
\$\begingroup\$

There are various research papers related to this topic and many methods have been tried.

Assuming you cannot use 35 cameras as mentioned above or a Hawk-eye camera based system, you may want to consider a radio transmitter placed inside the ball. This generates a signal that is in turn is picked up by antenna attached to the goalposts (see this patent document for more details).

There are some distinct advantages to this technology, the signal transmitted by the ball shouldn’t be significantly affected by weather, lighting or the presence of players. As long as the antenna receive the radio waves, a computer system can theoretically calculate the ball’s position. A disadvantage, is the necessity of placing a radio transmitter within the football. Firstly, the exact point from which the signal is transmitted will determine the position of the ball. This seems obvious, but if the transmitter moves within the ball, the system will deem the ball to be moving, even if it isn’t. This has significant implications when millimetre accuracy is demanded and the ball deforms significantly when struck.

Source: Here

Something similar is implemented in the iBall recently approved by FIFA. Inside the soccer ball are sensors that wrap around the ball in both directions.

The sensors inside the ball consist of a web of copper wire that uses induction, allowing communication with an antenna array that is mounted to the goal frame. The second that the entire ball has passed the goal line, the system sends a signal to a watch that the referee wears allowing the referee to know the goal was made.

Adidas are also developing Smart Ball technology. The ball has a computer chip inside that will relay information to the referee. This also makes use of cameras however. Cameras are placed at a certain angle so that it is clear when the ball crosses the goal line.

\$\endgroup\$
3
  • \$\begingroup\$ Thank you for your in-depth answer and providing some further reading material. I agree with the advantages of using signals as opposed to cameras! Thanks again. \$\endgroup\$
    – Sam
    Commented Feb 3, 2018 at 15:05
  • 1
    \$\begingroup\$ Glad this helps! Many methods have been tried and there is lots of research material out there. The main reason being World Cup goals being wrongly disallowed (or vice versa) that have put FIFA under some pressure to come up with a solution. \$\endgroup\$
    – rrz0
    Commented Feb 3, 2018 at 15:09
  • \$\begingroup\$ For sure! On a side note, do you know reliable sources for hardware (building the transmitter/receiver). The obvious constraint will be size and durability given it's going to be kicked about a lot, literally. \$\endgroup\$
    – Sam
    Commented Feb 3, 2018 at 15:15
0
\$\begingroup\$
What I am trying to achieve is the position of the ball as a goal is scored.

I think this can be solved quite reliably with four cameras.

Remember that a goal is scored the instant the whole of the ball passes out of the field. The edge of the field is defined as the outside of the lines, which is also supposed to be the back edge of the goal posts.

I haven't tried this, but my first reaction is to put two cameras on the back edge of each of the vertical goal posts. One maybe two feet from the bottom, the other two feet from the top. Remember that the goal opening is 8 feet tall.

These cameras would be looking right down the plane that is the edge of the field. They are therefore in the best position to see if a goal was actually scored, and if so, exactly when. Since the positions and the view angle for each pixel are known up front, it only takes two cameras compute a position. With four cameras, the goal keeper can be blocking the view from one side, and you still get the minimum two views you need from the other side.

\$\endgroup\$
1
  • \$\begingroup\$ Interesting approach in overcoming the view blocking of the goalkeeper. If using cameras, will camera quality need to be considered before purchasing. i.e. will this only work with cameras of a certain specification? \$\endgroup\$
    – Sam
    Commented Feb 3, 2018 at 15:22
0
\$\begingroup\$

Embedded sensors in the ball were tried and ruled out because of the lack of range resolution with near field sensors and rotation of the sensor. Therefore vision systems are the only reliable method.

Computations must be made seconds after an event is triggger to determine the 2D tangent surface of camera at the ball position near the goal line. When the intersection of each tangential surface using the computed ball centroid and it's circumference has crossed the inside of the goalpost plane a goal is detected and then the relative parallelogram computations result in where the XY intersection occurs relative to the goal aperture and thus the position of the ball crossing the goal as if it were viewed straight on.

Although 4 (2D) cameras may be sufficient , just as in GPS location, six cameras (6) are better set for situations with ambiguity, visual impairment low SNR etc. to improve error rate to acceptible levels.

As it happens I found out such a system has already been FIFA qualified "HAWKEYE"

http://www.fifa.com/about-fifa/news/y=2012/m=7/news=ifab-makes-three-unanimous-historic-decisions-1660541.html

A more recent photo, this time from the Hawk-eye site illustrates 7 overhead cameras with the ability to detect a goal with any 2 camera and position accuracy improves with the number of camera and quality of image. 1mm tolerance indicates pretty high resolution.

enter image description here enter image description here

Accuracy is 1mm crossing the goal line. Ball position accuracy is the size of the ball.

\$\endgroup\$
2
  • \$\begingroup\$ The bottom left camera in your picture looks like it would either get in the way of the AR, or be sometimes blocked by the AR. The same thing applies to the bottom right, but also with players taking a corner kick. \$\endgroup\$ Commented Feb 4, 2018 at 16:35
  • \$\begingroup\$ Agreed. so they must be overhead cameras. \$\endgroup\$ Commented Feb 4, 2018 at 18:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.