Towards a Virtual Coach for Boccia: Developing a Virtual Augmented Interaction based on a Boccia Simulator

Abstract

Disability can be a factor that leads to social exclusion. Considering that involvement in society is paramount for a person with disability, participation in sports can be a powerful tool for inclusion. Based on this premise, the authors propose an intelligent virtual coach for Boccia to encourage the practice of this sport on persons with disabilities, while promoting social inclusion and shortening the learning curve for individuals new to the sport by learning about game strategy. The envisioned virtual coach will rely on Artificial Intelligence models, thus requiring the creation of large datasets, namely for ball placement and throwing movement recommendations. To answer these problems, this work is focused on the development of a Boccia simulator. With this simulator, it is possible to generate artificial gameplay images and allow the user to control an avatar with body tracking. Gesture recognition was implemented with a state-machine, thus enabling the player to throw the ball, with customizable physics, by performing one of two different throwing movements. This functionality can allow the recording of data describing the body movement associated with the placement of the ball in a certain position within the virtual court, which is essential for the proposed recommendation system.

An Augmented System Based on Machine Learning for Boccia Assisted Gameplay

“…1). The view from the camera is used as input of a deep learning model in order to detect, in real-time, the Boccia balls within the camera Field-of-View (FOV) and sort them according to color (red, blue, or white), which was the focus of previous work [3], [16]. To compute the distance of each red and blue ball and the jack (white ball), the corresponding centroid coordinates of each ball within the camera FOV are used, allowing to map all balls and to compute the score for the current game situation.…”

“…Finally, the detection area located at the bottom of the centre area of the interface is where the predicted class of movement is displayed to the user via image. As previously mentioned, this project follows the work done by the research team in [3], and so the GUI Fig. 8 was developed in the scope of the same project.…”

“…It can be played individually or in teams, and the goal is to throw or roll a set of colored balls as close as possible to a target ball, known as the jack. A system using augmented reality to enhance the Boccia game experience, aiding novice players in comprehending key tactical and physical aspects of the sport, ultimately resulting in a more enjoyable learning experience was proposed by the research team in [3]. One of the features of the proposed framework concerns the detection of the key gestures that can be executed by the players during a Boccia game using the data from a wearable device.…”

“…Following this trend, the present work focus on the developments concerning the gestures recognition module of the framework developed by the research team and presented in [3]. In order to fulfill this goal, a dataset was created and a deep learning model based on a Convolution Neural Network (CNN) was trained.…”

A Magnetic Transducer for the Detection of the Fetal Engagement Level in Part-Task Trainers

“…Finally, a desktop/mobile application allows the simulator to be used by students and instructors in different modalities: training, auto-evaluation and exam mode [16], [17]. The visual feedback has been implemented in Unity3D, a cross-platform game engine widely used in gaming, filmmaking, architecture, engineering, robotics, and more generally non-game simulations and visualizations [18] - [21]. The existing software showed in real time the orientation of the fetus in the birth canal and highlighted the fontanels whenever the user touches them.…”