“Once you find a problem you are invested in, you have this constant obsession and it persists until you solve it,” said third year CS Ph.D. student Stephen Butler.
He feels many people are driven to find solutions because of this constant “itch”. He said, “You try to fall asleep, you go to eat your breakfast, and all the time it’s there, always sitting in the back of your mind. The only way to relieve pressure created by your obsession is to work on the problem. It doesn’t go away until you’ve developed a solution or presented the work to your peers and received their feedback. Then you sign off and can move on.”
Even so, Butler said researchers are almost always in a state of obsession because even before they publish their first solution, the scientist has already been developing new ideas and applications.
“This same drive to discover is consistent with the tech industry,” Butler said. “Every year the big tech companies are launching new products or improvements to their products. It is not so different from research, and you have to purposely draw a demarcation line when you’ve ‘done enough’ and can move on.”
Finding a problem that has been known for a long time but remained unsolved makes him very happy. His research focus is in the dynamical motion planning aspect of robotics. For example, the navigation pattern a robot vacuum uses to clean a room may seem erratic to a human, but its pre-set algorithm has some assurances of complete coverage for the size and shape of a room and the obstacles inside that space.
Unlike limited-motion robot vacuum cleaners, the systems Butler prefers are dynamic and have very high degrees of freedom (DOF). He said, “You’ve seen the DRC videos of robots falling down, right? We’re still really bad at planning for these types of systems and that’s an interesting problem for me. I take a cross-discipline approach to problems like this.”
Butler recently published a paper on how to convert desired robot motion paths into trajectories that map how the robot limbs and joints actually move. He used the illustration of moving a book between two surfaces.
“Let’s say the robot needs to pick up the book from a shelf and transfer it to a table. Each of the motors powering the robot arm need to be instructed when to engage and how much torque to use at the appropriate time. By projecting the desired path onto a parametric line, the algorithm can efficiently compute a precise map of time and force instructions for each motor which can be executed in the shortest amount of time,” he said.
Butler was able to craft his algorithm to take a generic path, convert it to a trajectory, and also maximized time efficiency for the calculation. Using his research, trajectory calculations for high-DOF, dynamical systems that might have taken a week to compute previously can now be ready in a matter of minutes with strong guarantees.
Before coming to Rice, Butler worked for Boeing as a mechanical engineer primarily assigned to NASA’s docking system (NDS) design. He said, “They wanted a soft capture system. Current docking technologies require space craft to either impact each other fairly hard, which is less safe than one would wish, or in the case of the ISS the spacecraft is grabbed by a robotic arm which is a slow, laborious process.”
His manager encouraged his interest in a graduate degree and helped Butler work out a schedule that allowed him to work full time on the NDS project while pursuing his M.S. “That lasted a year,” he said. “We successfully finished our critical design review and then I left Boeing to focus on my Ph.D. and my passion in artificial intelligence and robotics.”
The first time he declared his interest in the field was in elementary school. “My first grade teacher asked the ‘what do you want to be?’ question and I replied ‘a robotics engineer.’ My hobby was radio-controlled fliers. If you know what you’re doing, you can make a pizza box fly.”
Even though he was living in a small Texas town at the time, Butler’s interests in robotics and engineering were shaped and encouraged by his parents. His father was a Navy pilot and his mother – who had been one of only a couple women in the civil engineering department when she got her degree at the University of Texas – was teaching physics.
Twenty years later, Butler had completed the masters part of his graduate program and was progressing rapidly towards his Ph.D. He said, “When you have a firm goal in mind, you don’t get as distracted and that keeps you on track. I’m already working on another interesting project, but it has taken a while to develop the algorithms.”
His ultimate goal is to run a robotics company. “Robotics is one of the technologies that is on the cusp of being turned into actual products,” he said. “You can get more funding and momentum for your robotics research in the private industry than in the university setting.”
Butler is interested in finding solutions that can be applied to products now, even though his research is primarily on the theoretical side. He said, “The industry people I’m working with respect that it may take a year or two to develop the theory to the point where it can actually be applied to a product. To bring something new to the market, you have to spend the time in research first. Fortunately, there are many great government funding agencies releasing small business innovation research (SBIR) grants to fuel transformative research!”
Stephen Butler completed his M.S. in CS in 2017. His adviser was Lydia Kavraki.