虚拟现实技术阅读材料_A Steering Algorithm for Redirected Walking Using Reinforcement Learning.pdfVIP

IEEE TRANSACTIONS ON VISUALIZATION AND COMPUTER GRAPHICS, VOL. 26, NO. 5, MAY 2020 1955 A Steering Algorithm for Redirected Walking Using Reinforcement Learning Ryan R. Strauss, Raghuram Ramanujan, Andrew Becker, and Tabitha C. Peck Member, IEEE Fig. 1. Three different steering algorithms simulated on a real user’s path through a virtual environment. The physical tracking space is a square with 5.79 meter sides. The user’s virtual distance traveled was 6.4 meters with no redirection (left), 8.5 meters with steer-to-center (middle), and 9.2 meters with the Steering algorithm learned via Reinforcement Learning (SRL) (right). SRL resulted in a more arched path than steer-to-center, which allowed it to reach a greater virtual distance before colliding with the boundary of the tracking space. Gaps in the paths are a result of resetting that occurred when a user collided with a boundary of the tracking space, as we do not consider movement during the reset to be part of the virtual path. The user started in the bottom center of the environment and walked north. The passage of time is indicated by the light to dark transition. Abstract— Redirected Walking (RDW) steering algorithms have traditionally relied on human-engineered logic. However, recent advances in reinforcement learning (RL) have produced systems that surpass human performance on a variety of control tasks. This paper investigates the potential of using RL to develop a novel reactive steering algorithm for RDW. Our approach uses RL to train a deep neural network that directly prescribes the rotation, translation, and curvature gains to transform a virtual environment given a user’s position and orientation in the tracked