The goal of this work is to implement and analyze different methods that enable robots to navigate through crowded environments, such as those that may be encountered in shopping malls, road side walks, or industries. The two key methods considered here for planning a robot’s trajectory safe from moving and static obstacles are neural-network based estimation of a human’s trajectory and social forces method. We integrate these broad methods with other primitive shortest path-planning algorithms such as A* search to find a collision-free shortest path for the robotic agent through a crowded environment. To analyze the said techniques we use certain existing simulation environments and present a comparison between the approaches.

Report of the project I did with my collegue: get the PDF.

project repo