Ex Baidu deep learning chief sees future in AI
Horizon Robotics founder says cutting edge products will help fill vast future demand
Right before Chinese tech giant Baidu Inc started a road test of its self-driving car last year, Yu Kai, who was in charge of Baidu's autonomous driving business, decided to leave and established his own company, Horizon Robotics Inc.The goal was ambitious: to develop tailor-made chips and operating systems for artificial intelligence. The plan was formulated well before Google Inc's AI-powered AlphaGo defeated a world champion player of the ancient strategy game Go in March 2015.
"We are developing software solutions and hardware together in the hope of bringing more AI into our lives," Yu, also former head of Baidu's Institute of Deep Learning, said in an interview with Chinese tech website 36kr.com.
In April, the company raised an undisclosed amount of funding from Yuri Milner, the Russian investor who has bankrolled such internet giants as Facebook Inc and Alibaba Group Holdings Ltd.
According to Yu, Horizon Robotics aims to provide products that can support the application of AI in different industries. He said existing systems and chips can not offer enough computing and data processing power to drive AI-enabled products.
"By 2025, we will have AI inside almost everything," he said. "Most customer electronic devices will become robots in broad terms. They will perceive the environment, have interactions with humans, make decisions and have motion controls."
To date the Beijing-based firm has unveiled two versions of AI systems: the Anderson platform for smart homes and the Hugo platform for smart driving. Based on the two platforms, developers can design a variety of applications.
With the help of Anderson, home appliances such as refrigerators and TV sets can identify users through face recognition, "learn" from consumers' habits and then automatically make decisions and adjustments such as placing new orders online when there is no milk left in the refrigerator.
As for Hugo, it is essentially an advanced driver assistance system that can offer drivers real-time detection in pedestrian lanes and roads. It is robust and can work in poor weather conditions, including rain storms.
The company said it expects that the two systems will be able to support 1,000 types of electric products within 10 years. Yu said a leading global auto parts supplier, who he would not identify, had adopted the company's monocular sensing technology and they had signed a commercial contract.
Tan Tieniu, a member of the Chinese Academy of Science, said though the AI industry as a whole was still in its infancy, AlphaGo's historic victory proved how far AI could go once it was fully commercialized.
"But it is important to stay calm and sober-minded and put more effort into research and development of key technologies," Tan said.
He said the global AI market was worth some $127 billion in 2015 and would exceed $165 billion in the current year.
