通过重定向行走(Redirected walking, RDW)技术，人们可以在较小的物理追踪空间中以比较真实的体验来探索大型虚拟世界。重定向行走算法会通过不易察觉的调整来控制用户在物理空间中行走的轨迹，从而尽量减少用户与物理空间的碰撞。过往的研究主要集中于重定向算法的引导上，大致能被分为预测算法和反应式算法两类，大部分工作都在不可感知重定向增益的前提下进行研究。本文将用户行为和意图考虑到重定向行走算法中，为重定向行走算法提供了一些新的思路和方向。本文的主要贡献包含了以下两点：提出了一种基于历史用户行走数据的重定向行走算法。预测式算法在适用的情况下会有更好的效果，反应式算法在各种虚拟场景中都更加通用。而通过对历史的用户行走数据进行分析，即使是在相对自由的场景，也能够在一定程度上对用户的行走进行预测，这些预测信息将影响到算法的决策。算法从仿真的历史用户行走数据提取得到的行走热力图和加权有向图，以此作为用户在虚拟空间中行走的预测信息来提升重定向行走算法的效果，减少用户体验虚拟现实时的重置次数。实验表明，考虑了历史用户行走数据的方法在各种环境尺寸和障碍物布局中优于多种最先进的方法。提出了一种由用户意图驱动的重定向行走算法。在不同的场景下用户的感知阈值会有所变化，在虚拟空间中的某些场景下用户甚至愿意接受能被感知的重定向增益。对此，本文将虚拟空间按区域划分开来，每个区域内用户都能选择不同的重定向增益范围。本文设计了针对虚拟空间中不同区域存在不同重定向增益范围的重定向行走算法，该算法主要利用贪心的思想，根据用户当前重定向增益范围来选择对用户接下来的引导方式，实验表明该算法在各种虚拟空间重定向增益区域划分情况中优于绝大多数经典重定向算法。基于以上研究成果，本文研发了一个考虑用户行为和意图的重定向行走系统，设计了使用多段搜索来充分利用路径预测信息和重定向增益约束的算法，该算法在各种虚拟环境和物理空间的组合中能极大减少用户体验沉浸式虚拟现实时的体验中断次数，其效果优于多种最先进的方法。

With redirected walking technology, people can explore large virtual worlds with a more realistic experience in a smaller physically tracked space. The redirected walking algorithm controls the trajectory of the user's walking in physical space through subtle adjustments to minimize collisions between the user and the physical space. Past research has mainly focused on the guidance of redirection walking algorithms, which can be roughly divided into two categories: predictive algorithms and reactive algorithms. This paper takes user behavior and intention into account in the redirected walking algorithm, and provides some new ideas and directions for the redirected walking algorithm. The main contributions of this paper include the following four aspects:A redirected walking algorithm based on historical user walking data is proposed. Predictive algorithms work better where applicable, and reactive algorithms are more versatile in a variety of virtual scenarios. By analyzing the historical user walking data, even in a relatively free scene, the user's walking can be predicted to a certain extent, and these prediction information will affect the algorithm's decision-making. The walking heat map and weighted directed map extracted by the algorithm from the simulated historical user walking data are used as the prediction information of the user walking in the virtual space to improve the effect of the redirected walking algorithm and reduce the reset when the user experiences virtual reality. frequency. Experiments show that the method considering historical user walking data outperforms several state-of-the-art methods in various environment sizes and obstacle layouts.A redirected walking algorithm driven by user intent is proposed. The user's perception threshold will vary in different scenarios, and in some scenarios in the virtual space, the user is even willing to accept the perceived redirection gain. In this regard, this paper divides the virtual space into regions, and users in each region can choose different redirection gain ranges. This paper designs a redirected walking algorithm for different regions of the virtual space with different redirection gain ranges. The algorithm mainly uses the greedy idea to select the next guidance method for the user according to the user's current redirection gain range. Experiments show that the algorithm It outperforms most classical redirection algorithms in various virtual space redirection gain region divisions.Based on the above research results, this paper develops a redirected walking system that considers user behavior and intentions, and designs an algorithm that uses multi-segment search to make full use of path prediction information and redirection gain constraints. The algorithm is used in various virtual environments and physical spaces. The combination can greatly reduce the number of experience interruptions when users experience immersive virtual reality, and its effect is better than many state-of-the-art methods.