Session: 02-10-01: Session #1: Robotics and Automation in Advanced Manufacturing

Paper Number: 88658

88658 - Grid-Video Measurement Method for A-UGV’s Small Obstacle Avoidance Performance 

The factory logistics is changing into the autonomous era, and the Autonomous-Unmanned Ground Vehicle (A-UGV) can provide enhanced real time services in connection with other machines or information systems in factories. The services can be more helpful when A-UGVs have high level autonomy and are able to fully collaborate with manufacturing systems. The safety and performance of A-UGVs must be measured in detail under the factory environments, and ASTM Committee F45 has been developing standards for A-UGV performance measurement in various domains. Among the standards being developed, the performance of avoiding obstacles in factories needs to be managed carefully as it may cause severe damages to factory properties including A-UGVs themselves. Particularly, small obstacles that are un-/mis-detected are highly likely to with, so they must be considered more carefully when measuring A-UGV performance.

In this paper, the grid-video measurement method is proposed to measure the small obstacle avoidance performance of A-UGVs. As a first, this paper confirms the needs for defining the A-UGV performance for small obstacles through existing examples. Unfolded cardboard boxes and hinged door chocks are examples of common small obstacles existing in the manufacturing environment. The hinged door chock cannot typically be recognized by A-UGVs, and A-UGVs respond in varying ways to one or more stacked unfolded cardboard boxes. By measuring how A-UGVs respond to small objects in typical environments, performance clues are exposed about what users can expect from A-UGVs or how to operate A-UGVs. Based on these examples, A-UGV capabilities, when near small obstacles, have been summarized in terms of: sensing, object detect distance, travel and navigation, and collision.

 

Next, the grid-video measurement method is introduced as a low cost, standard method to measure the small obstacle avoidance performance of A-UGVs. The grid-video measurement method can measure the distance when interactions between the A-UGV and obstacle occur by using a grid around a small obstacle, capturing the grid using video, and then replaying the video to show grid-measured, A-UGV performance changes depending on the conditions. The method mainly deals with A-UGV specification, small obstacle and radial or square grid, and the environment as they all cause significant changes in the A-UGV performance.

 

This paper describes the concept, requirement, and procedure of the grid measurement method. Then, to verify the effectiveness of the grid-video measurement method, A-UGV experiments measured small obstacle avoidance performance. The measured performance is an A-UGV to obstacle distance when the A-UGV detects the obstacle. Two toy blocks with a 130 mm height and 15 mm depth were used as small obstacles, and widths were 31.5 mm and 63 mm. The A-UGV speed, the obstacle width, and brightness were used as experiment variables for A-UGV specification, small obstacle, and environment, respectively.

 

As a result of testing 30 times for each variable combination, a total of 240 tests, the A-UGV speed showed a large effect on the average distance, the obstacle width showed a small effect on the average distance and standard deviation, and the brightness showed a large effect on the standard deviation. Also, the success rate of finding an obstacle significantly changed depending on the combination of variables, and in the most difficult condition, the obstacle was not detected at all in 30 trials. Through this experiment, it was verified that the grid-video measurement method can be effectively used to measure the small obstacle avoidance performance of A-UGVs, and it is sufficient to be a low cost, minimal effort, standardized tool. This paper contributes to the marketing and use of A-UGVs by explaining the need to define and manage performance against small obstacles in operating A-UGVs and supports this notion by proposing an inexpensive measurement method. In addition, by quantitatively analyzing small obstacle avoidance performance upon A-UGV specification, obstacle sizing, and varying environmental conditions, this paper provides improved understanding and use of A-UGVs. The method and experimental results proposed in this paper will be used for ASTM F45 standard development.

Presenting Author: Soocheol Yoon National Institute of Standards and Technology

Presenting Author Biography: Dr. Soocheol Yoon is a research associate in the Intelligent Systems Division at NIST, and a post doctoral fellow in the Institute for Soft Matter at Georgetown University. His current work focuses on performance measurement systems for industrial autonomous vehicle and exoskeleton under ASTM F45 and F48 standard development technical committee. Dr. Yoon's research interests include A-UGVs' performance measurement under various industrial conditions such as obstacles, vehicle status, and operating environments. Dr. Yoon received his Ph.D. in industrial engineering at Pohang University of Science and Technology (POSTECH) in South Korea.

Authors:

Soocheol Yoon National Institute of Standards and Technology
Roger Bostelman Smart HLPR, LLC
Ann Virts National Institute of Standards and Technology