Flight of the Shuttlecock
Analyzing badminton shuttlecock trajectories in VR.
ShuttleSpace is an immersive analytics system that assists experts in analyzing trajectory data in badminton. The system visualizes stroke trajectories along with their usage and winning rates, and enables comparison between categories of trajectories.
The majority of existing systems for visualizing this kind of data present the trajectories in 2D court diagrams, making it difficult for experts to fully understand the situation on the court. ShuttleSpace enables experts to inspect these trajectories in 3D from the point of view of the players using VR. It also displays important 2D statistical data related to the trajectories which are often required by the experts. It does that by leveraging peripheral vision to combine the 2D and 3D visualizations. It also introduces VirtualStroke, an interaction imitating a badminton stroke using the VR controller to query the trajectories. This selection method is natural for the experts and enables them to efficiently interact with trajectories without using keyboard and mouse. In addition to VirtualStroke, it also supports other common interactions such as click, touch and gaze. Experts can develop effective game tactics using the insights gained from this system.
Let’s walk through a general use scenario of ShuttleSpace. The expert loads the player’s stroke data into ShuttleSpace. After wearing the VR headset, the expert is immersed in a full-size badminton court simulated by it. The trajectories are automatically grouped into five categories. It visualizes these categories by displaying their average trajectory with each category having a different color. Two semi-donut charts are visualized on the left and right sides of the expert’s FOV, presenting the usage and win rates of the categories. The smaller semi-donut chart at left presents the usage rate of techniques used in previous strokes.
It also visualizes the player’s trajectory meaning their start and end points on the court when they perform a stroke. For each category, it is shown as a polar coordinate on the ground and it’s centered at the average start point for the category. The direction of the coordinate follows the FOV of the expert. Two grid-based visualizations are attached on the left and right side, presenting the usage and win rate distributions along the distance between start and end points. The expert then selects a category by using the VirtualStroke interaction.
The system then enters a detailed view and unfolds the trajectories in this category. This view brings two grid-based visualizations which show the distributions of usage and winning rates along the vertical axis. This vertical axis could be the start point of shuttlecock trajectory, highest point of the trajectory, or the end point of shuttlecock trajectory. This is decided by the point at which the expert is looking at. This visualization helps them to understand the effect of these positions on the usage and win rates.
It displays these 2D statistical visualizations in the screen space so that they follow the expert’s movement and always appear within their FOV. This positioning also avoids them obstructing the trajectories.
The expert can also compare two trajectories. Their usage and win rates are visualized using the donut view giving the expert a summarized comparison.
To compare the trajectories at a deeper level, it uses a grouped grid-based design in Grid View, where each row is divided into two sub-rows, each presenting the data of one trajectory.
In the IEEE TVCG paper describing the work, the researchers discuss collaborating with four actual badminton experts to develop ShuttleSpace. An expert explored the lob strokes of Kento Momota, who is famous for his unpredictable style of play, to find his weakness. When viewing Momota’s trajectories, they discovered that a stroke category “ C2” is his most frequently used strokes, but has a low winning rate. They also discovered that this trajectory ends close to the back-right side of the court. Moreover, according to Grid View, Momota is more likely to lose the rally if the opponent returns the shuttle using a smash when it is higher than 2.5 meters. With this data, they came up with a tactic that after forcing Momota to perform a stroke of C2, the player should move back to the back-right side of the court and prepare to smash the shuttle at a place that is higher than 2.5 meters.
All experts gave overall positive feedback for the system. However, it was observed that wearing a VR headset is tiring and it also takes some time to get familiar with VR. They also noted that the legibility of text labels and legends needs improvement. They also provided some valuable feedback on the VirtualStroke interaction. They found it difficult to query trajectories which require high level stroke skills. Also, they suggested that the interaction could imitate haptic feedback on stroking the shuttle.
Research Paper
- Shuainan Ye, Zhutian Chen, Xiangtong Chu, Yifan Wang, Siwei Fu, Lejun Shen, Kun Zhou, Yingcai Wu. ShuttleSpace: Exploring and Analyzing Movement Trajectory in Immersive Visualization. IEEE VIS 2020.
