Hao-Kai Wen's Website



RASP: A Comprehensive System for Complicated Actions in Halfpipe Sports

Graceful actions performed by an athlete can be used to make an action sequence panorama (ASP). However, we seldom understand complex actions by ASPs.

It is action sports which involving fast and complex actions usually make it hard to be understood by the audience. Therefore, we chose inline (inline-skate) vert (halfpipe) as the target sport of our case study to see if ASP can assist users to understand complicated sports actions.

During this research, we found that the details of an action are hard to be presented in the ASP because actions in different temporal steps are overlapped together. To overcome this issue, an effective and natural recomposition method of ASP (RASP) is proposed. Since our recomposition method is developed based on the way that athletes developing their tricks, the usage of the proposed system is natural and even be unaware to users.

Besides, the developed enhancing tools, i.e., the halfpipe model and the action figure we used in user study can also be seen as an assisting learning tool for users, they provide an effective way for both professionals and amateurs to comprehend complicated actions in halfpipe sports.

Event Detection in Broadcasting Video for Halfpipe Sports

In this work, a low-cost and efficient system is proposed to automatically analyze the halfpipe (HP) sports videos.
Analyzing HP video is challenging because

  1. there are often more than one shots taken at different locations by shoulder-mount camcorders in a single trick segment, and
  2. the scenes in an HP video are motion blurred since players move fast.

Fortunately, by utilizing the HP court color coverage ratio in the scene and saliency object detection mechanism, we overcame the challenges. The proposed system segments video into trick level precisely.

Besides, a novel and efficient method for detecting the spin event is proposed on the basis of native motion vectors existing in a compressed video. Experimental results show that the proposed system is effective in recognizing the hard-to-be-detected spin events in HP videos.

Decore Testing of Wheels with Polyurethane Case and Polyester Core

Decore is a common and non-observable flaw in the polyurethane (PU) wheel that causes the case and the core of a wheel not adhere to each other firmly.

In this work, a non-destructive testing method was proposed to determine if a wheel is normal or abnormal. We installed an accelerometer on the wheel, hit it by the pendulum, and measured the impulse response. And, the defected wheel can be detected by the characteristics of the impulse response.

Traditionally, the decore in a wheel can be determined only when the wheel has broken down. In order to verify the proposed non-destructive testing method, the destructive testing method was also applied to the wheels. "Shear stress adhesive test" and "cyclic shear stress adhesive test" were performed by universal testing machine to define the adhesion of the core and the case of a wheel and to verify the proposed method.

The Robotic Arm Picking the Ball

The robotic arm is built by ourselves from mechanism design, manufacture, microcontroller programming, and moving path simulation. And we used only existing college resources.

Initially, balls are placed on the table.

  • The camera captures the image of the table and transmits it to the PC.
  • The PC calculates the position of the ball by the computer vision module.
  • The user chose the ball, she/he wants to pick.
  • The PC illustrates the moving path of the robotic arm picking the ball by 3D animation.
  • The robotic arm picks the ball.


The Sensing Motion Bike Light

A conventional bike light only brightens the center of the road. When cornering, it can barely brighten the upcoming road.

Polestar is a lighting device which solves the above issue. It senses the acceleration, determines the direction the bike turned to, and brightens the upcoming road.

In addition, Polestar also provides these functions:

  • lighting up the rear light while braking and
  • automatically turning on the light when riding into tunnel or in the dark.