Hand Gesture Based Gameplay with a Smoothie Maker Game Using Myo Armband

Serious games have the potential to guide the relearning process via encouraging and motivating meaningful interaction. This paper focused on assessing the feasibility of gameplay by performing hand gestures using an off the shelf myoelectric armband to make smoothies in a functional game. The game was designed in Unity3D and interfaced with the wireless Myo Armband as an input device for performing the tasks in game. Based on earlier work on feasibility of incorporating machine-learning based gesture recognition, cylindrical, spherical, and tripod grasps were incorporated into the game. Smoothie Maker game was designed with two versions Game-A & Game-B. Participants, (n=20), were randomly assigned to an AB or BA group which differs in order of gameplay for the two games. After playing each game, participants offered their insights using the Intrinsic Motivation Inventory (IMI). The results featured multiple parameters including score, time to pick, idle time, as well as gesture recognition accuracy for both game versions. Most outcomes indicated that games A and B did not have a statistically significant difference, but when comparing using gesture accuracy, the two game differed slightly with statistical significance. Analysis of the qualitative IMI survey did not provide a significant difference between the two game versions. Conclusions are drawn from our findings towards improving the games and their recognition accuracy, highlighted for our future work.