Development of a Bluetooth controller for mobile VR headsets

Abstract: Mobile virtual reality (VR) headsets have been becoming more and more popular. However, the cheapest headsets do not come with any controllers and the ones that do include controllers only uses sensors for rotation, not translational movement. This thesis project aims to develop a prototype of a Bluetooth connected controller for the mobile VR headsets. The controller is based on a MetamotionC board produced by mbientlab Inc., which comes with Bluetooth Low Energy (BLE), an ARM M4 microcontroller, an miniature inertial measurement unit (IMU) sensor (containing a 3-axis accelerometer, a 3-axis gyroscope, and a 3-axis magnetometer and a barometer), a thermometer and other sensors. The only sensors used in this project are the accelerometer, gyroscope, and magnetometer. As a finished prototype, the MetaMotionC is placed on a glove together with five Aruco markers; a 3D model of a hand intended to use as an avatar of the glove was made with Blender and MakeHuman; and a VR room to use the controller with was created in Unity. The 3D hand responds to rotational and translational movements via Bluetooth connection to the IMU sensor on the MetaMotionC. The smartphone camera is used to detect the glove's position with Aruco markers, and the 3D hand is moved to a corresponding location in the VR room. The OpenCV library is used for image processing. The sensor data is filtered with low-pass, median, and thresholding to improve the measurement accuracy. Zero velocity update is used to reset the drift of the integrated accelerations. To reduce the integration error, Romberg's method with a floating window is implemented in Matlab. However, it did not reduce the error enough to make a difference. Thus, the result was unreliable.