Hawkeye in sport

Advances in sport technology aren't just aiming to improve athlete's performance. Recently, we've seen the emergence of technologies designed to help officials enforce the rules of their particular sport. Probably, the best known of these is Hawkeye. Ginny Smith spoke to Luke Aggas, Director of Tennis at Hawkeye Innovations Ltd to find out how the system works.

Luke - It's a camera based system with the cameras located at the back of the arena. So, it's a totally non-invasive system. There's nothing put inside the lines of the courts or the tennis ball itself. It utilises some bespoke and advanced vision processing techniques to detect movement within the field of view of each of those cameras and then combines what it believes to be a tennis ball from each of those field of views to give you a 3D position relative to the court lines.

Ginny - So, Hawkeye is not just being used for tennis. It's used for cricket and it might be being brought in for football, but these sports have quite different balls. They're different colours and in the case of football, it's quite a lot bigger. How does that affect the technology?

Luke - We're very fortunate in terms of the general colour and the size of a tennis ball, so it's very consistent from event to event. In the cricket, where the system was first used, depending on the match or the test or the tour a different ball can be used in terms of the colour. A ball can discolour especially they start using the white ball and although people at home believe that the Hawkeye is automated, no-human intervention sort of system, in order to make sure that the system is tracking optimally under all conditions - be that the ball has changed colour, a shadow has come across the tennis court, or with all electrical equipment they'll be some moment where, something will need powering down and powering back up again. So, that's the reason why we stick up more cameras and there's some human intervention to make sure the system is working optimally throughout.

In the football, the ball will change between your optimal conditions through to, obviously, we know the use of the orange ball in snowy conditions. Equally, we're very, very used to the pattern as well as the colour changing on the ball between competitions and that's something that we're given clear direction on well in advance of those competitions, what ball they're going to use and we actually are given samples of that ball in order to fully test the system for robustness and a reliability point of view to make sure the system is working optimally under all conditions.

Ginny - So, with cricket, obviously, Hawkeye is used in leg before wicket calls where you're having to predict something that didn't actually happen because you're looking at whether if the leg hadn't been there, the ball would have hit the wicket. So, how do you go about predicting something like that that has so many variables?

Luke - The cameras that we use in cricket frame rates of those cameras are a lot greater than the tennis primarily because of that reason. So, we're trying to achieve as many frames as possible, using ultra-motion cameras running at 350 frames per second. Obviously, it's not easy to validate and test the system as something that effectively didn't happen. But obviously, in a similar way, to a predict development within the tennis from a frame by frame point of view, the system will then forward predict the fly path of the ball.

Ginny - You must have to take into account things like the texture of the surface and the direction of the wind. Is that all part of the software?

Luke - It is. We calculate the court pace in a tennis sense, the coefficient of restitution, coefficient of friction on the playing surfaces that we track upon. In the tennis, a lot of people say, "What about the wind?" We're not effectively predicting the ball flight within tennis. We're joining the dots together. We're capturing the flight of the ball and therefore, if there is a given lob or shot under windy conditions, it does do a sort of boomerang then we're just joining those dots together. So, that is all taken into account. We are capturing the actual video footage rather than what the system believes. In a predictive development within the cricket that is an unknown obviously in that short space of time between the ball bouncing and the ball contacting the player's pad and then passing the wicket. So, in that space of time, for wind to have a significant influence on the ball, it's extremely, extremely unlikely and obviously, the relative weight of the cricket ball is influenced by the wind that much less than a tennis ball.

Ginny - So, with football, you're using it as a line technology, so just to see whether a ball has passed the line and is a goal. Is there any way you could use it for other things perhaps to see if a foul was really a foul or if it was just a tackle, whether they actually touched the ball, that sort of thing?

Luke - As with all sports, we kind of create the technology for our customers and at the moment, goal-line technology is being the only specification in terms of any form of video replay and we've created that technology and hopefully fix that age-old problem. Saying that, we are very, very open to other suggestions if this has opened the door on using video to best officiate football, offsides being a clear one. Also there are technologies that we have provided in other sports, as well as the Olympic games, where we've taken in numerous video feeds from standard broadcast cameras, synchronised those cameras so that you can see what is happening in each of those field of views or in each of those cameras at that exact moment in time. So, to give rugby as an example, they obviously go to the officials to see whether the try scorer or supposed try scorer's foot had touched the line prior to the ball contacting in the floor. At the moment, they look at that on a camera by camera basis and the technology, the way we're currently various sports is to be able to synchronise that footage and to be able to see the output of that, all at the same time.