situation, waiting until the shot is almost completed before reacting. If this applies to ice hockey, we can look at the moment the puck leaves the stick as the ...
Ice Hockey Goaltenders’ Strategies, Reaction Times and Anticipation Times in Two- and Three-Dimensional Virtual Environments Hugh Tyreman1, J R Parker2, and Larry Katz 3 1
Member, Sport Medicine Centre, and Graduate Student, Faculty of Kinesiology, University of Calgary. Director, Digital Media Laboratory, Faculty of Fine Arts, and Professor, University of Calgary. 3 Professor and Director, Sport Technology Research Laboratory, Faculty of Kinesiology, University of Calgary 2
Abstract. Three-dimensional virtual environments are expensive to create and require specialized equipment compared with two-dimensional virtual environments, but either maybe be useful in improving reaction times, anticipation times, or strategies. This study used 10 novice and 9 expert ice hockey goaltenders. The participants were presented with 16 unique shots in two-dimensional video and again in three- dimensional video for a total of 32 shots in a realistic ice hockey goaltending situation. The order of presentation of the shots was random. The participants were asked to make saves on wrist, snap, slap, and backhand shots from both right and left handed shooters. The participants’ reaction times and anticipation times were tracked with Ascension Technology Corporation’s “Flock of Birds” tracking system. A flock of birds motion sensor was attached to the gloves measured positions and velocities in real time, which permitted the computation of reaction times and anticipation times. The differences in reaction and anticipation times were calculated from these measurements. All shots were high shots that needed catching or blocking. The differences in reaction and anticipation times were modeled by experience level, gender, shooter hand, shot side, participant handedness, dominate eye, shot type, goaltender style, experience in playing goal in other sports, and shot velocity.
Keywords: anticipation, virtual reality, strategy, response time, virtual video environments
1. Introduction This study was an exploration of the differences in reaction time and anticipation time in a virtual reality training environment. We placed novices and experts into either two-dimensional or three-dimensional virtual environments playing the role of ice hockey goaltenders. Working with ice hockey goaltenders, Miklanek and Scholzova (1999) used a standardized reaction test on a PC and showed a best reaction average of 193ms on the right hand and 203ms on the left hand. On a “complex” task which involved the choosing of the side and reacting, they measured a best right-hand time of 258.7ms and a best left-hand reaction time of 304.7ms. Unfortunately, no standard deviations were reported. Kuhn (1988), looking at penalty kicks in soccer in the German Bundesliga and the European Cup, suggested that there is open or closed loop strategies employed by penalty kickers and goaltenders. Is there a similar relationship between shooters and goaltenders in hockey? Savelsbergh, Williams, Van der Kamp, and Ward (2002) looked at goaltenders’ anticipation times in soccer defined as when the ball leaves the foot. They suggested that the expert goaltender maybe in a race situation, waiting until the shot is almost completed before reacting. If this applies to ice hockey, we can look at the moment the puck leaves the stick as the starting point for the anticipation time. Another explanation for analyzing reaction time, could be comparable to a boxer making a decision in order to be at a particular spot to block a punch or, if inaccurate, get hit. Looking at boxers, Ripoll, Kerlirzin, Stein, and Reine (1995) found no differences in reaction times between experts and novices, but in complex situations they did find a difference in accuracy.
In a study of tennis players, comparing live action, video, and point models researchers found that, with additional information (3D), experts improved in anticipation time, whereas novices took longer to make decisions (Shim, Carlton, Chow, & Woen-Sik, 2005). Based on this model, we would expect novices to react more slowly and experts more quickly in three-dimensional video environments. However, in twodimensional environments novices should react faster, and experts should react more slowly than they would in three-dimensional environments. The use of interactive virtual environments (handball) in the training of athletes has been observed to produce a similar physiological reaction to that of the actual game environments (Bideau et al., 2003). Technology has the potential to maximize the impact of training for the individual because modern systems can adapt to the objectives and needs of the athlete (Katz, Parker, Tyreman, Kopp, Levy, & Chang, 2006; Nigg, 2003). This study looked at reaction/anticipation times between 3D and 2D, by novice and expert in simulated environments. Creating simulated environments are expensive, especially 3D, so identifying the effectiveness of the interaction between environment and expertise may be of value from the perspective of performance and cost effectiveness.
2. Methods The training scheme being devised is as follows: a variety of different shots taken on a hockey goaltender from slightly different positions are recorded using a video camera positioned near where the goaltender would be standing. When played back on a large 12ft screen, it appears as if the viewer is standing in the net and an opposing player is shooting at them. Participant goaltenders stand in a hockey goal in the laboratory and their reaction times in response to the shots in the videos are measured. The video recordings were presented in standard 2D and in stereoscopic 3D mode, and both types were tested on the participants.
2.1. Subjects Volunteer male and female participants were recruited by poster, word of mouth, and email of the poster. The novices were largely students at the University of Calgary. The goaltenders were from the university, Rent a Goalie, and varsity teams.
2.2. Video Content Creation The video was captured by a twin-lens camera, storing each lens information in a different interlace to create a three-dimensional video. The shots were taken 2 male varsity players, one left-handed and one right-handed. The video was saved in DV format to maximize quality and editing ability. The twodimensional video was created by replicating the odd interlace over the even interlace, removing the stereoscopic effect from the video. The material was than transferred in DV format to a SONY DSRDR1000 video playback unit. The SONY DRS-DR1000 was controlled from a PC using custom software to synchronise with the data collection devices. Each shot was noted for the type of shot, shooters handedness, and shot location. The shots’ velocity were measure by Stalker ATS radar gun which is accurate to +/- 0.1 mph, has a speed range of 1 mph to 300 mph, and a range of 400ft with baseballs. The system was used indoors at a range of 20-40ft in an ice hockey rink, well within the performance characteristics of the system.
2.3. Test Environment A plastic ice hockey net, right and left handed hockey goaltender gloves, full sized straight bladed hockey goaltender and Velcro to attach the sensor to the gloves were the props in the experiment. The participants’ motion data is captured by Flock of Birds, using two sensors – one on each glove. The Flock of Birds captured data at 100 Hz, accurate to 0.01 inches in the x, y, and z dimensions in distance from the transmitter. The transmitter was located on the floor in the middle of the ice hockey goal. The video was projected on a 12ft screen, by a VREX VR-4200 video projector. The participants wore Model #60GX active display glasses, which were synched to the video projector by infrared. The experiment was controlled by a PC, with custom software to synchronise the video and motion capture.
2.4. Procedure The expert and novice participants were further divided into two groups. The shots were randomly assigned an order and broken in groups of 8 shots as a block. The first group saw the first block of eight three-dimensional shots first, then a second block of eight two-dimensional shots, then the first block of eight two-dimensional shots and, finally, the second block of eight three-dimensional shots. The second group saw the second block of eight two-dimensional shots first, then the first block of eight threedimensional shots, then the second block of eight three-dimensional shots and, finally, the first block of eight two-dimensional shots. This study was designed to check for learning effects and whether people learn at different rates depending on which video is presented first. • • • •
•
Participants completed a questionnaire on demographics and experience Participants performed a warm-up, followed by stretching. Participants had a training session of four shots, two in three-dimensional and two in twodimensional video to acclimatize them to the environment. Then the blocks of 8 shots were started using the following procedure: o 4-sec blank screen, allowing goaltender to rest o A 4-3-2-1-sec countdown appears and goaltender gets set o Shot starts o Goaltender reacts o Shot finishes and blank screen returns Cool down, research explained
3. Results The average age of the expert female goaltenders was 26.3 (7.0 45.6) years, the youngest female goaltender was 20 years, and the oldest was 35 years. The average age of the male goaltenders was 29.7 (23.7 35.6) years, the youngest male goaltender was 23 years, and the oldest was 38 years. The average age of the female novices was 27 (22.0, 32), the youngest female novice was 24 years, and the oldest was 31 years. The average male novices was 27 (18.4, 35.6) years, the youngest male novice was 19 years, and the oldest was 41 years. The female goaltending group consisted of 3 left hand catchers and zero right hand catchers. The male goaltending group consisted of 5 left hand catchers and one right hand catcher. The novice female group consisted of 3 left hand catcher and one right hand catcher. The novice male group consisted of 5 left hand and one right hand catcher.
3.1. Reaction Times differences in two and three dimensions The difference in reaction time will also be confounded by the different length of the shots. Since the shots do take a different amount of time to materialize from the start of the video clip, this will confuse the issue and make clarity much hard to achieve. An ANOVA of reaction time by dimension (F=0.08 p
