Future Tense

Can a Wall-Climbing Robot Teach Your Kid to Code?

A Harvard team hopes to fill the gap between demand for computer science education and the lack of qualified teachers.

Emily Harmon, a third grader at the Woodward School in Southborough, Massachusetts, puts Root the code-teaching robot through its paces, along with Raphael Cherney, research assistant for Harvard's Wyss Institute for Biologically Inspired Innovation.
Emily Harmon, a third-grader at the Woodward School, puts Root through its paces, along with Raphael Cherney, research assistant for Harvard’s Wyss Institute.

Amy Benford

This story was produced by the Hechinger Report, a nonprofit, independent news organization focused on inequality and innovation in education.

In May, a few hand-size, hexagonal robots took over a third-grade classroom in Southborough, Massachusetts. They climbed a whiteboard and drew all over it while flashing multicolored LEDs and chirping musically. All the while, they were teaching kids to code.

Meet Root—a robot being beta-tested by its creators at Harvard’s Wyss Institute for Biologically Inspired Engineering. The Wyss team hopes Root will soon roll into the gap between the growing enthusiasm for K-12 computer science and the lack of qualified teachers. With Root’s help, they claim, any teacher can become a computer science teacher.

“Root’s job is to celebrate the code you create by bringing it to life,” said Justin Werfel, a senior research scientist at Wyss. Root magnetically clings to whiteboards (most of which are metal-backed), where it acts out programs that students compose on iPads that are wirelessly linked to the robot.

Unlike other educational robots—such as Bee-Bots, Dash & Dot, and Lego Mindstorms—that are geared to a specific age range, Root is meant to span from prekindergarten to college. With black sides and a plain white top crossed with LEDs, Root is deliberately uncuddly and unadorned—a highly functional, sensor-packed box that can draw with a marker inserted in its middle.

“Bringing a really cute robot into your kindergarten classroom is great,” said Zivthan Dubrovsky, who heads the robotics group at Wyss. “But then you have to keep buying new robots as the kids get older. There’s all this cost, and the kids have to keep mastering new hardware.”

Root’s flexibility is based in its coding app, called Square, which has three tiers of complexity. In level one, users program with a picture- and block-based code that includes very few words. A curving arrow is the command to turn. A purple boomerang will put a loop in your code. Drag in a musical note and select the specific tone from a pop-up keyboard.

Root’s sensors can spot and react to what you’ve drawn on the whiteboard—following a black line, for instance, or stopping at a red one. The robot can also lift up and put down the marker to draw as it moves.

Level two of Square is also block-based, but it introduces more text and numeric variables for things like movement commands. Level three is text-based, JavaScript code. Students can toggle between the levels to translate their block commands into text code or vice versa.

Root grew from two strands of research in the lab of Harvard computer science professor Radhika Nagpal, a Wyss core faculty member. In the summer of 2011, a visiting undergraduate in Nagpal’s lab named Raphael Cherney made a magnetic robot. Cherney, now a Wyss research assistant, pegged his creation as a robotic whiteboard eraser that he called MAG-NEATO.

But when more senior colleagues saw the wall-climbing robot, they knew it was destined for more than cleanup duty, even if they didn’t know exactly what that destiny would be.

The answer crystallized when a postdoc in Nagpal’s lab, Michael Rubenstein, entered a 2012 African Robotics Network challenge for a low-cost educational robot. Rubenstein, now a Northwestern University computer science professor, won first place with the Affordable Education Robot, or AERobot, which cost $10 to build.

Like Root, the AERobot was designed to help kids learn to code and had optical sensors that allowed it to turn toward a light or follow a path drawn on a piece of paper. There was immediate and worldwide interest from educators for the AERobot (now licensed by a Chinese company called Seed Studio, which sells them for $19 each, or about $14 a piece for a bulk order).

The AERobot’s success persuaded the Wyss researchers that they should develop a whiteboard-climbing robot for the classroom. Plus, AERobot had limitations that they wanted to overcome with Root and its programming interface.

First, AERobot was aimed at middle school students. It was too complicated for younger kids and too limited for more advanced students. Second, it had to be plugged into a computer to upload its programmed instructions, so it was cumbersome to tweak and test changes to student code. Finally, despite being geared to beginning coders, AERobot required at least some technical acumen to teach and troubleshoot in class.

“You’re not going to be able to take a sixth-grade teacher who knows nothing about technology, give them the AERobot, and then walk away and have them be successful,” said Dubrovsky.

It’s a challenge faced by schools everywhere—a push for coding and technology literacy but a shortage of computer science teachers. Earlier this year, the Obama administration proposed $4 billion to states, including $100 million directly to school districts, to retrain teachers as computer science instructors.

“We’re taking a different approach,” said Dubrovsky. “We are making a robot that will work out of the box and that teachers can use immediately. If they can download an app, then they can use the robot to help teach coding.”

The Wyss researchers brought three Root prototypes for last month’s test drive with the third-graders at the Woodward School in Southborough. They started with a few demonstrations, such as programming Root to turn in tandem with the iPad’s accelerometer (an internal sensor that tells the device when it’s being tilted), so that kids could “drive” Root by steering with their iPads. Then, they had Root follow a simple game board and programmed it to flash colored lights and play notes when it reached designated squares. Finally, the children all tried to program Root to draw their first initials, which most of them couldn’t do on the first try.

“There was a lot of trial and error and problem-solving. It was amazing listening to their conversations and to watch these kids,” said Amy Benford, Woodward’s technology integration specialist. “And when they were successful, their attitude was, ‘Look at what I just created!’ ”

The Wyss team is looking for more schools to beta-test Root prototypes this fall. The robot is not yet for sale, but the Wyss team estimates it will cost about $200. That’s 10 times more than the AERobot, Dubrovsky admitted, and assumes the school already has iPads. But, he argued, one Root up on a whiteboard can cover a lot more students.

Root’s developers hope to extend the robot’s social reach by creating a website where kids can show off their code and teachers can swap lesson plans.

“Longer term, we’d like there to be that kind of sharing,” said Dubrovsky. “We want Root to help turn learning to code into a social experience.”

Future Tense is a collaboration among Arizona State University, New America, and Slate. Future Tense explores the ways emerging technologies affect society, policy, and culture. To read more, follow us on Twitter and sign up for our weekly newsletter.