The US military is one of the largest engineering service entities in the world; however, due to the nature of its workforce, it does not have a large pool of service personnel with years of on-the-job experience. Often young and inexperienced warfighters need technologies to help them quickly and efficiently learn assembly and maintenance skills on engineered products. Augmented reality (AR) shows promise delivering work instructions in a user centered format, improving first time assembly accuracy over traditional 2D interfaces. However, building an effective AR instruction system faces many challenges. This paper addresses a critical one: navigation around an assembly area using AR. The work in this paper compares three different navigation methods: points of interest (POI), path planning gates, and a 3D directed arrow. The POI interface consisted of a simplified 2D top down view of a work cell showing stations and the user's position in relation to those stations (like a radar map). The path planning gates used a technique adopted by some flight simulators: a series of yellow squares marking the path from the user's position to the area of interest. The third method was a floating three-dimensional arrow, which rotated in the scene to point towards the participant's next area of interest. To test these three navigation methods studies were conducted. Participants were asked to assemble a mock aircraft wing twice. Their performance was recorded via subjective self-report and by objective sensor-collected data. The dependent variables included assembly time, errors, movements within the assembly area, net-promoter score and time spent looking at instructions. Findings indicated that for navigating around a work area, the gates interface was the best, resulting in lower completion times, decreased time spent looking at the instructions, and ratings by users, compared with the other interface methods.