The second match between 1.5S and StarPro began like the first. The 1.5S hero made a beeline to the island and began filling up on golf balls. But this time, instead of focusing their fire on the hero as it climbed to the island, StarPro hunted for 1.5S rovers instead. They managed to take out two of them before the 1.5S hero filled up on golf balls and made it off the island and back into combat.
The 1.5S and StarPro teams operated their robots from a small booth on the raised stage behind the battlefield. The audience could see the engineers’ faces projected onto giant screens above each team’s hut, but the players controlling the robots could not see out; the window in front of them was made of an opaque glass. This ensured that they had to navigate using the camera feed from their rovers and hero, in addition to the high-level view from their team’s drone camera.
The 1.5S hero tried again to destroy the enemy base, but it didn’t have the damage boost from the computer vision test. It ran out of ammunition, and had to head back to the island. As it returned to the battlefield, the StarPro hero drove underneath it, trapping it in limbo over the obstacle course, leaving its back wheels spinning helplessly a few inches off the ground. The crowd went wild. It was the perfect counter to the highly technical hero that had so far dominated the competition. Everything now came down to the third and final match.
The games had been running for three days straight at this point, with battles raging for 12 hours a day. The hallway surrounding the arena had been made into a makeshift camps for competitors. Engineers soldered broken parts back together, sending small plumes of smoke into the air. Others slept on the ground beside their battered bots, spare styrofoam tucked under their heads for pillows. It was clear the students were willing to work extraordinarily hard, but they were doing it on their own terms. Over the last decade young Chinese have become increasingly unwilling to embrace the brutal grind of life on the factory floor, to take the kind of manufacturing industry jobs that gave rise to Shenzhen and laid the foundation for DJI.
There was a circular irony here. To compensate for rising wages and an aging workforce, many factories are looking to automate their production, turning to the same robotic technologies at the heart of this competition. President Xi Jinping has called for a “robot revolution” and promised 200 billion in subsidies to Shenzhen’s regional province for this transition. Many young, well educated Chinese will find work creating, controlling, and tending to these autonomous laborers at companies like DJI. Many students saw RoboMasters as a chance to move beyond the rote curriculum of their university and prepare for the more creative future.
RoboMasters began in 2013 as an internal competition, a chance for the engineers to blow off steam while still working on technology core to DJI’s business. It was small scale, held inside DJI’s office with a makeshift course. The second year, it was integrated with a summer camp DJI hosted for college-age engineers interning at the company. Last year, it opened to the public, and teams from universities across China, as well as a few teams from abroad, signed up. It began experimenting with a more lavish production, building out the battlefield with lights, music, and live announcers. It was a hit, widely covered by Chinese media, and so this year DJI went even bigger.
The company says it spent around $9 million on the 2016 tournament, although several employees told me privately that number was closer to $15 million. Months before the finals in Shenzhen, employees fanned out across the country to help organize regional contests between 228 teams. DJI engineers produced custom components for the thousands of fighting units that competed; the company even commissioned an anime series, set to air on Chinese TV this fall, about a nerdy teen who finds his calling, and his courage, behind the wheel of a rover.
Why all the effort and expense for a competition that doesn’t, at least directly, help sell any drones? The simplest answer is recruitment. As DJI forges a path forward in hardware development, it must tussle with international giants like Baidu and Didi, Uber, and Amazon, all of whom want top talent in robotics, computer vision, and autonomous navigation. DJI is also setting itself up to be the default robotics platform. Almost every major robotics program at a Chinese university uses DJI’s infrastructure to educate their students. In effect, a rising generation of engineers is being trained to work on DJI products, and to think of the company’s equipment as the gold standard.
But RoboMasters is also a passion project for Frank Wang, DJI’s founder and CEO. The 36-year-old is now worth billions, and has proven himself as a brilliant engineer and a ruthless manager. But in many ways the competition is a reflection of a simple truth about Wang: he loves playing with robots: building them, flying them, and watching them fight.
A large LCD screen on the front of the SkyWorth Semiconductor building plays a short anime film on loop. In it, a young boy falls in love with a remote control helicopter. But as he grows up, the drudgery of college and an office job choke his passion. A suffocating storm of papers envelops him, dragging him down to hellish landscape. He stands at the edge of a cliff, unsure how to go on. And then he makes the leap of faith, breaking free of his suit and tie, falling towards a lake of fire. At the last second he catches hold of a small white drone, which saves him from his plunge, and carries him aloft.
The inspiration for the cartoon is Frank Wang, who as a little boy loved the illustrated series 动脑筋爷爷 — loosely translated as “brainstorming grandpa.” The cartoon starred an elderly McGuyver type who helped kids to get out of jams through feats of clever engineering. Wang’s favorite episode featured a little red helicopter, and he went on to collect model aircraft and remote control copters. Growing up in a middle-class family in Hangzhou, he dreamt of attending a prestigious technical college in the US like Stanford or MIT, but didn’t quite have the grades. Instead, he studied electronic engineering at the Hong Kong University of Science & Technology instead.
Wang was a mediocre student in college as well, but he used his time at school to work on his passion: his senior thesis was building a flight controller for a miniature helicopter. That project barely got off the ground, but it impressed his professor, who encouraged him to try graduate school and helped him find a scholarship. In 2006 Wang left academia for good, moving to Shenzhen and using what was left of scholarship money to start a new company, DJI.
Located on the 21st floor, Wang’s office commands a broad view of Shenzhen’s ever changing skyline. It’s adorned with plastic replicas of fighter jets and origami Phantom drones. Toy mechs line one bookshelf, and plastic figurines of vintage propeller planes are arranged carefully on his desk. He displays the usual CEO reading material — The Art of War, Hatching Twitter, Steve Jobs, Ayn Rand — but also highly technical textbooks on subjects like superalloys, radio navigation, flow induced vibrations, and high temperature coatings.
Dressed in a white button down and dark slacks, Wang sported a soul patch and large glasses, which added a slightly hip air to his otherwise boyish enthusiasm. Though Wang’s team insisted our conversation focused on RoboMasters, and that he would only speak in Mandarin, Wang’s enthusiasm soon got the better of him. After a few minutes, he dropped the Mandarin entirely and spoke in English, pushing the conversation to a faster pace, and roving far and wide across subjects beyond the competition.
I asked what he felt the purpose of RoboMasters was. “For engineers, they do not have a stage, a competition, to become loved by lots of people, to show their wisdom, show their precision.” The tournament was designed to make stars out of nerdy college students, and in doing so boost interest in the field. “If we can put engineering and entertainment together, not only can they entertain, they can educate lots of people.” Not all would win, but even those who lost could go on to be engineers, innovators, and entrepreneurs, a side effect Wang felt was “good for society.”
RoboMasters is a war game, and Wang repeatedly emphasized the importance of aggression, a trait he believed was fundamental to DJI. “Very aggressive people are concentrated here, without the heritage of the bad culture. So the new things can grow in this kind of environment.” Wang contrasted the the practical and hypercompetitive nature of Shenzhen to the bourgeois pleasures of Shanghai or the bureaucratic power of Beijing. “So maybe it’s the spirit, the future of the Chinese spirit, is what I’m thinking.”
Aggression was a central part of Wang’s approach to innovation at DJI. A former employee who worked at DJI for years told me it was a “very burnout, six days a week minimum, competition based,” environment. RoboMasters, employee said said, was a fitting entrance exam for DJI. “Pitting teams against each other and having one win is how product development works inside of DJI.” Wang didn’t try to hide this. “Sometimes a smart decision will make a lot of people unhappy,” he told me. “If we let the company employees elect the leader I would never be elected, because I am too tough.” What works in Shenzhen, however, may not be as successful around the world. DJI has opened offices in Europe and the United States, but several employees said turnover has been high.
As to exactly what product DJI might build next, Wang was coy. “I think vision is really the key to enable new applications for robots. That is why we want to integrate it into our competition. And this is also the future of our company.” DJI drones use computer vision to sense and avoid obstacles as they fly. The drones, and DJI’s handheld camera, the Osmo, can also recognize and track subjects. Tap on the athlete or a race car you want in your shot, and the device will keep them in frame without you moving a muscle. “We first use the vision in the drone, but later we can expand to other applications, like autonomous driving, agriculture, autonomous cherry picking, this kind of thing. Lots of human, very labor intensive things can be replaced by cheap, vision enabled robots.”
The crowd was jubilant before the third and final match between 1.5S and StarPro. It was the first time any team had given 1.5S serious competition, and losing the final round meant the team would be eliminated from the tournament. The stands were full of families, many with young children who stared, mouths agape, at the giant robots projected onto screens above the battlefield, and screamed with glee whenever a unit toppled over, spilling plastic marbles across the turf. Film crews from local TV news were on hand to interview players, and young bloggers posted updates from their phones to the Chinese equivalents of Facebook, Twitter, and YouTube.
The match opened slowly. Both teams danced around the middle, moving warily, like heavyweight boxers cautious about stepping into the range of knockout punch. At the five minute mark both teams exchanged shots at close range, losing two rovers apiece. 1.5S placed one of its rovers to take the computer vision test, but was driven off by enemy fire before it could finish. Slowly the teams whittled down each other’s infantry until only the two hero bots remained.
StarPro gave chase. Its hero, less than half the size of the massive 1.5S robot, seemed to have its opponent on the ropes. 1.5S took a half dozen shots to its backside, and a nervous, confused murmur swept across the stadium. The two heroes came tumbling over a hill and crashed into a pedestal, rubbing paint like Nascar drivers jockeying for space. And then, with just over a minute on the clock, the 1.5S hero finally turned to fight. Four quick shots from its golf ball cannon and the StarPro hero went dark.
A day later, 1.5S went on to win the competition for the second year in a row — several of its team members were subsequently hired by DJI. I spoke with Junru Chen, the team leader of 1.5S, who had spent 10 months preparing for the championship. His shirt was soaked in sweat and pieces of gold confetti from the trophy celebration were sprinkled throughout his hair. Now that it was finally over, “We are very excited,” he said “but also relieved.”
The goal of RoboMasters, in his mind, was to look for people who were “technology addicts”, and to test them under the most extreme conditions. “It takes the form of competitions to unleash our potential,” he told me. What would he build with that potential I asked? What future did he envision? He said the goal was to integrate robots more deeply into the lives of average people, the Jetson’s fantasy of robotic butlers and maids.”To design machines which help people achieve some goals, first the robots need to listen and see,” Chen said. “Computer vision is a very important field for us to to achieve a smart lifestyle in the future.”
Right now RoboMasters lets DJI sift through some of China’s top engineering talent, but the company is hoping to expand the tournament around the globe. I chatted with Betty Vogeley, a student the University of Washington, the only team from outside of Asia to compete in RoboMasters to date. “We haven’t been doing so well, as far as winning any games,” she said with a laugh. “We’re known as the team whose wheels fall off.” Vogeley has been competing in robotics events across the US for the last six years, but had never experienced anything close to this challenge. “This is everything combined. It calls for so many diverse talents, there is just nothing like it.” She was eager to play host to qualifying tournament that would give US teams a chance to practice before the finals. “To make it competitive, we need to bring a regional tournament to North America.”
I caught up with Liu Zi Yi, from the Xi’an team. He and his classmates had traveled two days by train to arrive at the competition, because flying was too expensive. And they had sunk their own savings into the parts needed for these robots, even borrowing from friends and family. But he didn’t seem too upset by the loss. The contestants at RoboMasters saw Shenzhen and DJI as way to make a break with the past, and to secure a good future in a rapidly changing economy. “The city I am from, it is very traditional, very conservative,” said Liu. Everyone was very focused on following the rote, academic curriculum, on the safe career path. “If the past of the city is good enough, they can keep this. They say, this is enough. For a city as young as Shenzhen, they have nothing to lose.”
It seemed unlikely that he would find a job with DJI following his team’s performance, but Liu was eager to try his own luck as an entrepreneur. Wang’s story had clearly seduced many of the young men and women here. They were eager to work for him, or to follow his example, trying to build companies that thrived by innovating faster than the competition, not copying or undercutting them. “DJI changed the image of Chinese companies in the mind of foreigners,” said Liu. “It is the first, but it won’t be the last.”