Smart Wearhouse Human Robot Interaction

Samuel Falzone

Authors:  Samuel Falzone and Boyi Hu

Faculty Mentor:  Boyi Hu

College: Herbert Wertheim College of Engineering

Abstract

As of today, more than 10% of warehouses make uses of automated machinery to maintain and manipulate stock. Recent developments in automation technology have increased productivity for robots exponentially. However, these developments were made without proper testing of the impact automated machinery has on human workers creating a possibly false understanding of the necessary restraints. In response, a human-in-the-loop experiment was facilitated to simulate a smart warehouse setting where human and robot work in tandem. During the experiment the human completes a series of tasks including picking and assembly, transferring between two stations. This experiment tests three modes of human robot interaction: no robot, an empty robot, and a full payload robot. The robot simulates typical warehouse procedures with autonomous path planning and collision avoidance technology. The study takes each path the robot makes and records it in reference to the action of the participant, and notes the types and occurrence rates of direct interaction for each mode of interaction. Eight participants were recorded and analyzed by two individual researchers to present an accurate representation of the effects of automated technology on a human worker. Results from this pilot study can be used as design guidance for future collaborative robots.

Poster Pitch

Click the video below to view the student's poster pitch.

Poster

Click the image to enlarge.
0 0 vote
Presenter Rating
18 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Samuel Falzone
Samuel Falzone (@guest_402)
1 year ago

Test?

Samuel Falzone
Samuel Falzone (@guest_754)
Reply to  Samuel Falzone
1 year ago

Yup, I’m here and standing by for comments or questions. Feel free to leave a reply.
I also have a Zoom meeting going just incase so feel free to join that too!

https://ufl.zoom.us/j/521072385

Wayne Giang
Wayne Giang (@guest_970)
1 year ago

Hi Samuel,

Very interesting work. How were participants trained about the functionality and capabilities of the warehouse robots?

Samuel Falzone
Samuel Falzone (@guest_1134)
1 year ago

Hey! So participants had a training session that introduced them to how the robot would be moving and functioning before the recorded trails

Jeffrey Chen
Jeffrey Chen (@guest_3350)
1 year ago

Hey Sam,

Great poster! I was wondering how changing the time or speed of the robot traveling through its path may impact the distribution of interaction.

Samuel Falzone
Samuel Falzone (@guest_3798)
Reply to  Jeffrey Chen
1 year ago

Hey Jeff! Great Question!
So that Is exactly what we will be looking into in further studies. As mentioned on the poster there were a noticeable difference between some of the trials, We coined this difference as synchronized and unsynchronized. In further studies we will be testing small modifications made to the robot’s control scheme that have the potential to effect the interaction rates. Potential changes include manipulating the robot’s speed, delay times, or even the distance that the robot records its enviorments with.

Thanks!

Gabriel Streitmatter
Gabriel Streitmatter (@guest_4120)
1 year ago

Hey Samuel. Great work! Valuable information gathered here, I was wondering what pathfinding algorithm/techniques you used for your robot. Also, how was the robot sensing the environment?

Samuel Falzone
Samuel Falzone (@guest_5394)
Reply to  Gabriel Streitmatter
1 year ago

Hey Gabriel,
Sorry for the delay. From my understanding of the algorithm (A programming outsider), we have two algorithms that control the moments of the robot.
The global path planning is controlled using Dijkstra’s Algorithm to find the minimal distance path with the known map.
The local path planning is controlled by the Trajectory rollout Algorithm to simulate multiple trajectories based the current positioning of the robot.
The use of the Trajectory Rollout Algorithm was determined because it has been used successfully during a wide variety of tests that focus on other HRI enviorments.

Let me know if you have any more questions. Thanks!

Boyi Hu
Boyi Hu (@guest_4992)
1 year ago

Nice presentation, Sam! One question, by introducing robots in sharing the space work space with human workers, it seems humans will have substantial increased stress and mental workload, what are some of the barriers out there?

Samuel Falzone
Samuel Falzone (@guest_6860)
Reply to  Boyi Hu
1 year ago

Hey Dr. Hu!
Thanks for your question. Currently the largest barrier for implementing the robot in tandem with the human is the lack on information out there with specific focus on a warehouse setting. The warehouse setting creates a variety of specific demands. For example, in such an enviorments the human would constantly lose sight of the robot as it maneuvers through the warehouse. Also the increase load that the robot would have to be responsible for demands some specific considerations that need to be addressed before the worker can have an acceptable mental work load.

Haozhe Zhang
Haozhe Zhang (@guest_5070)
1 year ago

Hi,
Such an exciting work! I am doing some relevant work and this will surely inspire me.

Samuel Falzone
Samuel Falzone (@guest_7004)
Reply to  Haozhe Zhang
1 year ago

Thanks! Thats always an important out come with pilot studies like this, so glad to help!

Yuhao Chen
Yuhao Chen (@guest_5640)
1 year ago

Hi Samuel, fantastic work! During your experiment, did you notice any differences in human behavior affected by the size (empty vs full load) of the robot? Intuitively, working with a full load robot makes participants more anxious, and it might result in their behavior changes, such as tending to stay further from the robot and spending more time searching the robot’s location.

Samuel Falzone
Samuel Falzone (@guest_6344)
Reply to  Yuhao Chen
1 year ago

Hey Yuhao!
So to answer your question, it actually varied based on the participant. The participants that showed lower or no signs of searching for the robot in early trials with an empty robot, tended to show equal or less signs of signs for searching with a full robot. On the other hand, in the participants that showed the highest tendency to search for the empty robot, the full robot resulted in a large increase of “locational searching.” That was the general trend however, some participants showed a lower tendency to search for the full robot, presumably because it’s location was easier to determine with out actively searching for it.

Dr. Donnelly
Dr. Donnelly (@guest_6274)
1 year ago

Very good presentation, including relevance of project. Good use of graphics -all should be labeled. Nice work.

Samuel Falzone
Samuel Falzone (@guest_7074)
Reply to  Dr. Donnelly
1 year ago

Thanks so much Dr Donnelly, I really appreciate all of your support!

Yue Luo
Yue Luo (@guest_6286)
1 year ago

Hey Samuel. Great work! The poster looks neat and information in it are very valuable.

Samuel Falzone
Samuel Falzone (@guest_7190)
1 year ago

Thank you all for some great questions and responses! I belive that the Symposium has ended so I am signing off of the page and Zoom Chat.
If you have any further questions feel free to email me : sfalzone1@ufl.edu