Everyone has a smartphone, but within 10 years, everyone will also have a robot. So says Andy Xu, who wants Hexa, his company's six-legged, artificial intelligence robot, to usher in the personal robotics revolution.
"Twenty years ago, it was websites, then it was mobile apps. The next phase of technological innovation is robotics," says Xu, chief operating officer of Vincross, a three-year-old robotics startup in Beijing.
Hexa is a hexapod, which in plain English means that it is six-legged. It has 19 motors, sensors, A.I. software, and agile legs that can traverse and navigate different types of terrain and even dance. Xu says Hexa eventually could act as a companion for elderly people, similar to a pet. It has image, video, face, and voice recognition, and it may eventually be smart--its machine-learning algorithms help it learn new skills through trial and error.
Hexa's mobility is modeled after insects', says Xu. Vincross founder Tianqi Sun was a bioengineer before he launched Vincross in 2014 and raised $7 million from GGV Capital and ZhenFund. Sun wants Hexa, and Vincross's next generation of robots, to be considered artificial life forms, melding deep neural networks and machine learning. "We see robots as living beings," Sun told the publisher Tech in Asia.
Hexa is expected to go on presale for about $600 in a Kickstarter campaign this month. Developers can buy the robot now for about $1,000 to beta test it, and its retail price will be around $1,000 in 2018, according to Xu. Hexa has only a few applications, but Vincross is hoping developers will build apps, or what Vincross calls "skills," says Xu, to create a commercial version expected in 2018. Vincross opened its operating system, the A.I.-powered Mind OS, to universities and robotics labs across the world. Vincross is calling it a "human mentor" program--it is looking for more than 2,000 volunteers to write code for the robot so that it can learn on its own.
For example, Arash Tavakoli, who is pursuing a PhD in machine learning at the Imperial College London, is developing machine learning algorithms to help Hexa teach itself to walk and navigate its environment on its own. The algorithms process data from Hexa's camera and sensors and help it make decisions.
"I am not writing a program to make Hexa conduct a specific behavior, but rather providing a reward system to encourage it to keep trying new maneuvers until it successfully navigates its environment," says Tavakoli. "By trial and error, it will learn a policy for certain tasks."
Xu says Hexa has an easy-to-code operating system. In a few months, he predicts, the Hexa "skill store" in the Hexa mobile app will be filled with hundreds of applications.