Random-Forest-Based Initializer for Real-time Optimization-based 3D Motion Tracking Problems

Many motion tracking systems require solving inverse problem to compute the tracking result from original sensor measurements, such as images from cameras and signals from receivers. For real-time motion tracking, such typical solutions as the Gauss-Newton method for solving their inverse problems need an initial value to optimize the cost function through iterations. A powerful initializer is crucial to generate a proper initial value for every time instance and, for achieving continuous accurate tracking without errors and rapid tracking recovery even when it is temporally interrupted. An improper initial value easily causes optimization divergence, and cannot always lead to reasonable solutions. Therefore, we propose a new initializer based on random-forest to obtain proper initial values for efficient real-time inverse problem computation. Our method trains a random-forest model with varied massive inputs and corresponding outputs and uses it as an initializer for runtime optimization. As an instance, we apply our initializer to IM3D, which is a real-time magnetic 3D motion tracking system with multiple tiny, identifiable, wireless, occlusion-free passive markers (LC coils). During run-time, a proper initial value is obtained from the initializer based on sensor measurements, and the system computes each position of the actual markers and poses by solving the inverse problem through an optimization process in real-time. We conduct four experiments to evaluate reliability and performance of the initializer. Compared with traditional or naive initializers (i.e., using a static value or random values), our results show that our proposed method provides recovery from tracking loss in a wider range of tracking space, and the entire process (initialization and optimization) can run in real-time.