Dempster-Shafer's Approach for Autonomous Virtual Agent Navigation in Virtual Environments.

This paper presents a solution for the behavioural
animation of autonomous virtual agent navigation in virtual environments, which focus on using Dempster-Shafers Theory of Evidence in developing visual sensor for virtual agent. The role of the visual sensor is to capture the information about the virtual environment or identifies which part of an obstacle can be seen from the position of the virtual agent. This information is require for virtual agent to
coordinate navigation in virtual environment. The virtual agent uses fuzzy controller as a navigation system and Fuzzy for the action selection method. The result clearly demonstrates the path produced is reasonably smooth even though there is some sharp turn and also still not diverted too far from the potential shortest path. This had indicated the benefit of our method, where more reliable and accurate paths produced during navigation task.