Collision Prediction in a walkable VR environment

Master Thesis

About This Topic

The topic concerns highly immersive VR systems (HMD + head tracking in large scale) where multiple users can simultaneously explore virtual environments by walking. In non-shared scenarios where player do not see each other (for example, if they are on different floors in a virtual building) there appears a risk of collisions between player. Such dangerous collision situations should be prevented, preferably without the players noticing anything.

The first step of collision prevention is predicting possible future collisions. Such predictions can be made by extrapolating players' paths in a known virtual environment.

The specific task for the Master thesis is to develop a collision prediction algorithm for a constraint maze-like virtual environment. The algorithm can then be tested on pre-recorded paths. The working environment is Unity 3D, although Unreal can also be used if preferred.