Planning redirection techniques for optimal free walking experience using model predictive control

RedirectedWalking (RDW) is a technique that allows exploring immersive virtual environments by real walking in a small physical room. RDW employs so-called redirection techniques (RETs) to manipulate the user´s real world trajectory in such a way that he remains within the boundaries of the physical room. Different RETs were suggested and evaluated in the past. In addition, steering algorithms were proposed that apply a limited set of RETs to redirect a user away from the physical room´s boundaries. Within this paper, a generalized approach to planning and applying RETs is presented. It is capable of dynamically selecting suitable RETs and also controlling parameters like their strengths. The problem of steering a user in a small physical room using RETs is formulated as an optimal control problem. This allows applying an efficient probabilistic planning algorithm to maximize the free walking experience. The proposed algorithm uses a map of the virtual environment to continuously determine the optimal RET that has to be applied next. The suggested algorithm is evaluated within a user study and compared to a state-of-the-art steering algorithm. Results show that for the given virtual environment, it is able to reduce the number of collisions with the room boundaries by 41 % and furthermore reduces the amount of applied redirections significantly.