Using Mobile AR for Rapid Feasibility Analysis for Deployment of Robots: A Usability Study with Non-Expert Users

TL;DR

This paper tackles the challenge of quickly assessing the feasibility of deploying robotic arms without requiring expert-level tools. It introduces a mobile AR-based workflow that decomposes reachability analysis into six steps and leverages a cloud-based path planner to provide rapid results. Through an expert usability study and a quantitative study with 22 non-experts, the authors demonstrate substantial time reductions (roughly threefold) and generally maintained cognitive load and usability across task complexities, while identifying practical interface improvements. The approach has clear practical implications for SME adoption of automation by lowering cost and expertise barriers, though it remains complementary to, not a replacement for, full production-ready planning and optimization.

Abstract

Automating a production line with robotic arms is a complex, demanding task that requires not only substantial resources but also a deep understanding of the automated processes and available technologies and tools. Expert integrators must consider factors such as placement, payload, and robot reach requirements to determine the feasibility of automation. Ideally, such considerations are based on a detailed digital simulation developed before any hardware is deployed. However, this process is often time-consuming and challenging. To simplify these processes, we introduce a much simpler method for the feasibility analysis of robotic arms' reachability, designed for non-experts. We implement this method through a mobile, sensing-based prototype tool. The two-step experimental evaluation included the expert user study results, which helped us identify the difficulty levels of various deployment scenarios and refine the initial prototype. The results of the subsequent quantitative study with 22 non-expert participants utilizing both scenarios indicate that users could complete both simple and complex feasibility analyses in under ten minutes, exhibiting similar cognitive loads and high engagement. Overall, the results suggest that the tool was well-received and rated as highly usable, thereby showing a new path for changing the ease of feasibility analysis for automation.

Figures (5)

• Figure 1: Evaluating the feasibility of the machine loading automation.
• Figure 2: Proposed method workflow: a) workspace definition using bounding box; b) choice of robot model and placement; c) point cloud generation; d) point cloud processing: outlier removal, downsampling, brush eraser, sponge eraser, object addition; e) interaction area markup; f) path planning. Steps a-e) require user input and are completed on the user's device. Step f) is triggered by providing inputs from steps b), c) and e) and generates reachability results in the cloud. Best practices to obtain good scan is "free play": The user can always return to the previous step, make the bounding box larger or smaller as needed, move around to get a different perspective and use the tools to enhance the point cloud.
• Figure 3: Chart of the difficulty level of (A) and (B) tasks on 5-point Likert-like scale, as assessed before and after using our prototype with whiskers denoting standard deviation error.
• Figure 4: Flow chart of quantitative user study design.
• Figure 5: Cumulative results of TLX survey of tasks (A) and (B). The weighted average is calculated using pairwise comparisons. Lower scores are better (0--low demand/good performance, 100--high demand/bad performance).