SUMMARY

In dynamic and unstructured environments such as military zones, construction sites, and factories, workers have an increased risk of colliding with dangerous obstacles while engaging in their daily jobs. This can cause decreased worker productivity and more conservative planning by the human. In such concentrated sites, we can augment a person’s environmental awareness through motion planning and reduce the risk of worker related injuries. Previous work on navigation for humans has approached the problem in two different ways. The first is a local planning approach, which uses cues to tell the human exactly where to dodge an upcoming collision. While this first approach does address imminent threats effectively, using a less reactive method can augment a person’s environmental awareness beyond the instance of a collision. The second technique is a global path planner that connects a walkable path between different points of interest in an area and is seen in GPS or maps applications. A real-time global path planner can determine a possible route, but cannot directly map human movement to ensure a safe, collision-free one. Thus, I propose the design of an algorithm for human-based motion planning that integrates global and local planners, while accounting for the preferred walking dynamics of the human. This method could provide valuable insights for workplaces seeking to develop operator safety strategies, increase task completion speeds, or predict human path trajectories in cases of human-robot interaction. To achieve this, we will build a realistic environment with static and dynamic obstacles to test on human participants in virtual reality. This research aims to validate a real-time motion planner as humans attempt to maneuver in unstructured, dangerous environments.