A while back I interviewed Dr. Stuart Miller, the International Tennis Federation's (ITF) Technical Manager. At that time, Dr. Miller mentioned there were several new automated line calling systems on the market. The Cyclops system (used to call balls near the service line - in or out) had been utilized for years at major tournaments. While the ITF had no immediate plans to investigate such systems, Dr. Miller was interested in understanding the accuracy of all of these automated line calling systems.
I have always wondered how those systems worked. Several of the developers/producers of these systems made presentations at the Tennis Science and Technology Congress this past year.
My purpose here is not to question whether these systems should be used. I have just been curious to know how these systems work and thought I would share what I learned. I had a lot of questions and during this fact finding process, determined these questions were shared by others.
In some instances I realized these questions/concerns
actually defined system limitations or characterized areas
of improvement for these systems.
How would you design a line calling system? What does such a system need to accomplish? Clearly it can be difficult and a challenge for a linesperson to judge a ball's split second impact close to the service line. As well, a player's line of sight and visual perspective can prevent them from correctly assessing if a ball is in or out.
When a ball strikes a court, it compresses, deforms and spreads, becoming more oval in shape. Consider a serve that just strikes the outer edge of the service line. It is possible for the back of the ball to "overhang" and touch the service line. Yet, from the receiver's perspective and angle,
it may look like the ball was long and did not touch the service line.
The impact mark on some soft surfaces, like clay, can actually be larger than the actual "footprint" the ball makes on initial impact. On some surfaces like grass, the ball can slide after it strikes the court. How would you differentiate between the ball striking a line versus a player's body, racquet, an insect or a few rain sprinkles striking the a court line? Obviously, there are many considerations an automated line calling system must address.
You have probably seen (or heard) the Cyclops system which audibly calls balls in or out at the service line of the US Open as well as other tournaments. First developed in 1979, it was originally an upside-down periscope. The periscope lowered the operator's (which could be a linesman) line of vision down to the court surface. However, it took two people to operate the system and was still dependent on the operator's vision.
The Cyclops system evolved to a five infrared beam electronic system and has been used commercially for 22 years. The beams are emitted from transmitters and received by monitoring boxes which are located 12-17 feet from the sidelines. The beams run parallel to the service line with transmitter and receiver on opposite sides of the court.
One beam (the master beam) is positioned over the service line and four beams are positioned outside the service line at 5.1 inch intervals. When the ball crosses a beam and interrupts the signal, the system triggers a light on a handset or panel and, if desired for a fault, an audible sound. When a ball hits the service line, (and therefore is "in") a green or yellow light is triggered, while the four "fault" beams trigger one red light. It is possible for a good ball ("in") to cross the master beam and also cross the other fault beams. However the system has been designed to inhibit the fault triggers once the master beam has been triggered. However, if the ball misses the master beam (misses the service line and is out) it breaks the fault beams, a beep is heard and a red light activated. These beams are narrow enough to be interrupted by the ball but broad enough not to be triggered by an insect or small flying debris. Nevertheless, the system is only used on the service line in singles matches.
Another similar system, by JRM Engineering Ltd., has asserted that there can be certain ball angles which may not be detected with master beam systems due to the placement of the beam. This system uses a larger array of beams versus just a master beam and may provide even better accuracy.
However, Paul DelGrego, president of
FastCAM Replay LLC and DEL Imaging Systems LLC first told me about Cyclops several years ago when we were working together on a high speed camera project at the US Open. (DelGrego's team is the group responsible for the "MacCam" which displays high-speed footage of close and controversial line calls.) A comparison using those high speed cameras and the outcome of Cyclops was conducted and the results consistently matched.
One of the most interesting systems has been developed by Signal Processing Systems Inc. in Sudbury, MA. This electronic line calling system uses contact sensors physically embedded in the tennis lines. The sensor looks like a co-axial cable (a long ribbon with integrated insulated wires embedded in it). The ribbon is only .2 millimeters thick and can be completely covered with the coating material used to cover hard courts. For clay courts the sensors can be installed on the underside of a clay court tennis tape.
When the ball hits close to the line, auditory signals are sent to earphones worn by the line umpires. This system is intended to assist the umpire in close call situations and can be used on any line. Although players may step on a line and trigger the system, stepping on the line sounds more like the crunching sound of a footstep in snow versus the sharp thump of the ball. Cyclops has only been used on the service line for singles play, since it can be safe to assume the receiver will not step through the service line during the actual serve. This sensor system, however, can be set up on any lines, both singles and doubles. The system allows lines which are not to be monitored to be "shut off." The earphones are typically wireless.
A different technology is used by Auto-Ref Inc. in Toronto, Canada. This system couples four high-speed digital cameras with computer software which track, determine the trajectory and map the impact point of the ball.
Two sets of two cameras monitor each half of the court. Each camera is connected by a digital video cable to a separate computer. There is also a separate main computer. The computers are connected to one another via wireless technology. This system also has an operator controlled audible sound which designates that a ball is out. The system has several functions: it charts and recognizes a 3D map of the court in computer memory including lines and thickness; tracks the ball as it comes over the net and creates a 3D trajectory; predicts where the ball will strike the court; and determines if the ball is in or out. Additionally, the system has to differentiate the ball's motion from player or audience
movement. When the ball is blocked from the cameras view, the system predicts from the information it has where the ball will drop.
I was impressed with the diversity of technology used in these systems as well as the specification and depiction of scenarios these systems attempt to resolve. These innovators and others will undoubtedly continue to address and improve accuracy and consistency of automated line calling in tennis and in other sports.
Happy New Year .. Until Next Month ... Jani
