Hello! I’m Andre Cleaver, or you can call me Dre. I was born and raised in San Antonio, TX, and I am currently in the 3rd year of my Joint Ph.D. program in Mechanical Engineering and Human-Robot Interaction (HRI).
I will be posting a weekly blog post to fulfill my requirements for the Human-Machine Design course here at Tufts University, and since I am interested in Augmented Reality (AR), expect the majority of my posts to include AR devices and their role in HRI. In addition, I’ve added a Portfolio section to present my projects in a more professional manner.
The topic for my first blog post is Human-Machine Systems and Automation. As a student working with AR devices and robots, obviously this is an important aspect to me. Any given product will always have new ways to improve how a human-user uses the device. In my case, I’m exploring how much visual information an AR-device such as the Microsoft Hololens or Smartphone should be delivered to a user. For those unfamiliar with AR, you’ve probably seen it before in games and online shopping (Pokemon Go, Wayfair, Snapchat). These are all wonderful uses of AR, but the real power of AR in my opinion is the translation of robot/machine language to visuals that any human can interpret.
Take for example a Roomba robot, if you want to know where it plans to go next, you can open its path planner algorithm and you’ll probably see a scrollable terminal window full of numbers, words, and symbols that only the roboticist can understand. With an AR device, we can take that entire code and replace it with a simple line rendered on the ground to inform a human the future trajectory of the robot. Cool right?! But it’s still not straight forward. There’s a whole art to how these visuals should look, when these visuals should appear, and how much and how long do you show them before it becomes overwhelming. To figure this out, we need to test with humans.
As for automation, I am also writing algorithms that can influence both the robot’s and human’s actions. In the same Roomba scenario, a robot can have built-in sensors (cameras, bump sensors) to support its object avoidance algorithm and therefore replan its course of action in which the user can see through AR. As a result, the human and see the robot’s path and decide to replan their route or not change anything and expect the robot to change its trajectory. How the robot behaves and what I decided for the robot to show humans can have drawbacks.
I always recall the post, “The Code I’m Still Ashamed Of,” by Bill Sourour that tells an incident where a programmer built a website for a drug company that directs a visitor to take a quiz that outputs a recommended drug as a form of treatment that is based on their answers. Turns out that no matter how you answer the quiz, it will always recommend the client’s drug as the best possible treatment. In short, a young girl had killed herself who had taken the drug, possibly someone who had taken the quiz at one point. Although anyone can interpret this incident in several ways, it’s understandable that the creator of the website feels they are responsible.
I think about this post nowadays every time I’m coding for any project. I say to myself, “What is the worst-case scenario?” and try to ensure that I address those cases by stress-testing the code. I’m in a field where humans are participants, so any of my contributions to the field will have an impact on people, big or small. So far, I’ve done a couple of human-participant studies and all I can say is that humans can be interesting when they encounter a robot. Some people have never seen a robot and decide to stop and stare at them, some treat robots as if they are people (talk to them), and some fear of the robot take over and want nothing to do with them. I’ve been in the field for a couple of years now, so I have a reasonable expectation and mental model of a robot. Therefore, I try to get some friends who all differ in levels of coding to test out my code, and I note their reactions and comments and try to make changes accordingly. This act of Stress-Testing should be a norm for any Human-Machine Design Project.
With the emergence of new robots and technology, it is always important to keep in mind how you design these gadgets and how people will use and interact with them. The “worst-case scenario” is my mentality in hopes to reveal any flaws with my system.
I just read that article you mentioned “The Code I’m Still Ashamed Of” and what an interesting read. How sneaky the concept of 1st targeting young girls for your product and then giving them the illusion that by answering these specific questions an authority would ethically choose the right medicine for them. As you mentioned above, no matter what they answered on the quiz it still pointed them at the same drug every time. The fact that someone later committed suicide while on the drug is just terrible. I give the author props for leaving the company after he learned this. What makes this so nauseating is that technically it’s not illegal but it brushes up so close to that line it’s just so very sad.
Hello Dre, I really like the concept of visualizing information for easier consumption of information for people that may not be professionals in the augmented reality or robotics profession like myself. I’m excited to see the products that’ll come out of this field in the near future. I also appreciated your mention of the article “The Code I’m Ashamed of” which really highlights the grave consequences of decisions in the world of computer science and technology which as you mentioned is not unique on its own but strictly tied with humanity. I think your “worst case scenario mentality” and testing of your code under stress conditions is critical to proactive prevention of issues in the future.