If one were to enter room D14 at Mahan Hall, perhaps an aspiring student would be humbled by the setting. What was once a relatively spacious storage room became a makeshift laboratory for Class of 2021 Cadets Andres Sayed, Sara Scales, Austin Fox and Jafar Stone. This cadet team underwent a complex process of building Warehouse Autonomous Robots, from concept to completion, within the span of the academic year leading up to Projects Day on April 29 at the U.S. Military Academy.
Concurrently, to streamline their facilities, the Iowa Army Ammunition Plant sought a contemporary solution to the manufacturing process. Thus, with guidance from Lt. Col. Steve Crews, an assistant professor in the Department of Civil and Mechanical Engineering, the four cadets began the Warehouse Autonomous Robots capstone project to assist in the plant’s modernization.
“We wanted to optimize the efficiency through automation and transportation between buildings at the Iowa Army Ammunition Plant,” Stone said. “So we really wanted to include ideas of modularity, as well as graceful degradation to improve the overall efficiency and reliability of the warehouse system.”
The team’s efforts paid off as cadets, staff and faculty gathered in the Dean’s Conference room during Projects Day to recognize the Warehouse Autonomous Robots project with the 2021 Scott R. Clark Award for its demonstration of innovative problem-solving methods that plans to improve the Army Warfighter.
Brig. Gen. Cindy Jebb, Dean of the Academic Board, presented the award on behalf of the Superintendent, Lt. Gen. Darryl Williams.
“When you are downrange, you’ll be unloading ammunition and it will be coming from a warehouse somewhere. Maybe, the ammunition you receive will be manufactured from the robots that you designed,” Jebb said as she praised the cadets for their efforts. “This accomplishment is great. It’s good to see the list getting longer.”
This award, named after Scott Clark, USMA Class of 1985 graduate, honors a Projects Day effort that demonstrates an innovative approach to solving a problem of direct application to the Army.
The list of past Clark awardees included innovative projects such as Last-Mile Optical Communications, a laser-based computer network; the Soldier Power Augmentation Readiness kit, a battery charging system; the M-249 Squad Automatic Weapon reduce recoil, and the Raspberry Pi Android security system module.
Like the innovative projects of the past, kick-starting the development of autonomous robots involved many trial and error attempts. One of the first trials came with designing the laboratory to resemble a small-scale warehouse. The team also mimicked the automated process of a warehouse to get accurate data readings the ammunition plant could use. Following this, they used a variation of mathematical equations to develop robotic, one differential drive vehicle prototypes reusing assets from pervious projects, Crews said.
“For the cadet’s education to be worthwhile, they need to overcome the complexities of building a robot, if they spent their entire semester or their entire year just building a robot, I don’t consider that much of an accomplishment,” Crews said. “The accomplishment is when they can actually make that robot do something right. The accomplishment is having that robot actually serve a purpose.”
The process was stagnant in the beginning. The team had an idea of what they wanted the robots to look like and what function they would serve, however, the mathematical equations they used at the time only allowed them to develop basic prototypes rather than definitive products, Sayed said.
“Initially, a lot of this stuff was put together ‘franken-style,’” Sayed added. “We used leftover parts from other projects and came upon using these heavy (rectangular) modules and we set up aluminum 50 bars (in the front and back of it). We also used 3-Dimensional printed wheels that resembled bicycle tires and when we used our X-Box controllers to move the robots, the wheels would wobble in a warped sort of fashion.”
Despite the many setbacks the team would encounter, iteration would be the key to achieving the results they desired. Furthermore, in less than five months since the project began, four robots were developed with the help of the independent studies research from Class of 2022 Cadet Katherine King and Class of 2023 Cadets Alvin Ye and Alexander DiSalvo, respectively.
Algebraic equations originating from the Jacobi method started filling the blackboards, thick power cords snaked across the wooden floor and plugged into the power supplies of the three robot carts.
As the team received more viable resources and the research progressed, they eventually got their hands on a programming language and numeric computing system called MATLAB, which many institutions use in the realm of academia and industry. Coding, machine learning, communications and control systems are among the wide variety of uses MATLAB provides, Sayed said.
“Initially, we already knew what we wanted the products to look like, but the equation we had then versus the equation we have now were totally different,” Sayed said two months ago as the team started making significant breakthroughs with their research. “Another aspect of the research to consider is the more complex the equations became the more advanced the robot parts needed to be. The 3-D printed tires were eventually recycled and ‘mecanum’ wheels were now installed onto the robot allowing for more fluid mobility.”
The capstone team started their process with two robot carts and the robotic arm, however, King decided to do her independent studies project on robotics and contributed to the team by building an additional robot named the ‘Omni-Tank.’ The Omni-Tank was uniquely designed with its triangular frame, an omni-directional base and omni-wheels mounted on the three ends of each module, Crews said.
“Imagine you’re trying to parallel park your car. We all know it can be a tedious process. You have to do certain maneuvers and then back up carefully in the right angle in order to get the car safely into the space. With the omni-tank that King built, all you need to do is slide on in,” Crews said. “From concept to reality, it took about a week or two for King to build her robot.”
Ye and DiSalvo’s independent studies project focused on developing a battery management system that would allow the robots to be powered without the use of a power cord.
And yet, Crews would be the first to tell you that, “I know nothing about batteries.”
Ye and DiSalvo conducted their own personal research on what components were needed to develop such a battery.
“(Ye and DiSalvo) came to the department saying, ‘hey, we want to work on smart batteries.’” Crews said. “They visited Lt. Col. James Bluman, the Director of the Center for Innovation and Engineering, to pitch the idea and Bluman sent an email out to a bunch of faculty and it said something to the effect of, ‘hey we got two yucks — they’re exploring the realm of smart batteries. Is anyone interested in this research and would this help you in any way?”
Ye and DiSalvo wanted to work with smart integrated batteries from Tesla, specifically, and Crews knew he could aid them because the equipment they were using to build the robots was similar in concept to Tesla self-driving technology. Crews added it took him about an hour to convince the department heads that the robots the cadets were building needed the smart batteries.
He managed this by explaining that essentially, the robots are autonomous vehicles that require a consistent and powerful battery source. Hence, the batteries have to talk to the robot code they develop, which means if the battery goes idle, then the robot goes idle and vice versa.
With his thorough explanation, Crews got the thumbs up and the development for the battery began.
As Projects Day neared, constructive synergy continued to abound the room. An array of eight cameras were mounted high around the nooks and corners of the walls in the lab. The cameras served as a motion capture system for position location and orientation, informing the team where all the robots were.
“The cameras are communicating with our task manager, and the task manager is essentially like a global positioning system coming down and saying here’s where everything is,” Crews said.
Meanwhile, Fox focused on the path-planning technology, which detects the space and corridors that allow the robot to orient itself in the environment it works in and drives. Stone worked on controlling and moving the robots and making sure they would effectively communicate with each other as they carried on with their task. Once Stone had control of the robots, Fox would integrate the path-planning process that would allow the robots to see the world they are moving in.
“You can see these waypoints on the screen I coded for the robot. These waypoints are my plan for movement,” Stone said as he displayed the code sequences on seven different data files on his laptop.
Ideally, the robots would communicate with each other as they navigate the room delivering and collecting items from the robotic arm through a predestined pathway that has been coded into their modules, learning to avoid collision as they all move in a concerted manner.
Meanwhile, Scales worked diligently on the six-degree robotic arm that was mounted onto a mobile desk. Her laptop sat on a metal stool chair as she carefully sifted through the coded sequences, teaching the arm to grab an object from a small bin and place it in another bin.
“In the beginning of the semester, I came into this research project with no robotics experience like the rest of my peers. I was tasked with working on the robotic arm and I eventually learned how to control it,” Scales said. “In the second semester, I was still learning how to use the arm in more complex ways. In terms of receiving a major in robotics, I believe mechanical engineering is one of those branches where you can really work with the hardware.”
There were instances when the robotic arm would remain idle, sinking down slowly on its own due to its lack of weight distribution control. As a consequence, Scales had to consider a term called ‘gravity compensation.’
“Just like your arm, or just like a heavy arm, in general, it’s going to want to sag a little bit and that’s what kept happening when the robotic arm would sink when it was idle,” Crews said. “You need to add a little bit of torque in every joint to hold it up firmly in the right position and that is called a ‘feed forward torque,’ which is happening at each of the robots joints and gets it to stay still instead of sinking and, basically, when it’s in gravity compensation mode, you can move the arm around as if gravity wasn’t even there.”
Crews added that for the robot arm to achieve this, the robot arm needed to contextualize the pieces that make up its mass and where they are in relation to the world. The robot arm needed to know where its elbow was while moving to not collide with anything.
“Let’s say that you went to combat, and you came back, and you lost an arm. Imagine they attached this robotic arm to you at the shoulder. Now imagine the robot’s shoulder, down to its elbow and then its wrist. Now, how would you need to coordinate your robotic arm and yourself, so it doesn't hit you as you try to control it,” Crews explained. “The robot arm needs to have a sense of itself, relative to you so that you can properly control it. Therefore, it needs to have a sense of itself relative to the items around it in this lab.”
The cadets worked meticulously for the remaining two-and -a-half months they had left leading up to Projects Day. After spending two semesters writing research notes, organizing, coding and developing the four robots, the team finally presented their capstone project virtually to organizations and guests with interest in autonomous robotics.
During the virtual conference, the honorable Bruce Jette, USMA Class of 1976 graduate and the former U.S. Assistant Secretary of the Army for Acquisition, Logistics and Technology, expressed his excitement and delight in the hard work and laborious effort the team put into this project.
“I just want to say that you guys did a great job. If you were a defense contractor, this would have cost somewhere in the ballpark of about $20 million to get done, so you’re very efficient. Congratulations. You guys have a future if you want to go into industry at some point, but not yet because we really need you in the Army,” Jette said. “The limited application of some of our technologies, as far as what’s possible in those ammunition plants in all of our logistics facilities, really needs some attention. Your success in such a short amount of time, I think will significantly help that. And I’m going to do my best to make sure the right people know what you did.”
Social Sharing