By | Dazhuang Lv
Reported by AI-Drive
AI-Drive Note: The autonomous driving industry is not just about the few leading giants!
Recently, Israeli AI startup Hailo Technologies completed a Series A funding round, raising $12.5 million. This funding will be used for the development of deep learning chips to help edge computing devices perform real-time processing of high-resolution sensor data.
According to Markets and Markets, the market size for edge computing is expected to reach $6.72 billion by 2020. Meanwhile, IC Insights also indicated that global sales of automotive integrated circuits will exceed $42.9 billion in 2021. Additionally, let’s not forget the entire automotive industry’s pursuit of ADAS systems, where just providing computing power for deep learning algorithms can be incredibly profitable.
In a 2017 report on autonomous driving, McKinsey stated that the deployment of ADAS systems increased from 90 million units in 2014 to 140 million units in 2016.
“Both autonomous driving and ADAS systems require low latency, so integrating convolutional neural networks into onboard hardware is quite necessary,” explained IEEE Fellow Tom Coughlin. “Only fast and well-trained deep learning systems can make decisions in the blink of an eye like human drivers, helping vehicles avoid accidents, which qualifies them as competent autonomous driving systems.”
The new deep learning chips developed by Hailo will be used to run AI applications, and in the future, they will also enter markets such as drones, smart home devices, AR/VR platforms, and wearables. However, Hailo’s primary focus remains on the mobility market, with autonomous vehicles being the top priority. Chip samples are expected to officially debut in the first half of 2019.
“We believe that current computing solutions are severely resource-constrained and cannot meet the needs of large-scale applications of neural networks,” said Hailo CEO Orr Danon. “In our view, the key shortcomings are deeply embedded in the computer architecture. Existing architectures are still designed for traditional rule-based software, and without our technology, they simply cannot handle cutting-edge deep learning applications. After all, we need to move away from large data centers and ensure that edge computing devices achieve a balance in power consumption and cost. We believe this technology will bring a wealth of interesting new use cases, and the automotive industry will benefit first.”
“Hailo is building a new processor architecture that will significantly enhance the efficiency of deep learning tasks,” Danon added.
Danon pointed out that to process full HD camera data in real-time using deep learning algorithms, a data center-level GPU running at full capacity is required, and just one sensor can consume as much as 100 watts.
“During the R&D phase, such power consumption is acceptable since the testing vehicle’s trunk is filled with large computing equipment,” Danon explained. “With our technology, a few watts of power can accomplish the same tasks, paving the way for large-scale deployment. Both ADAS systems and future autonomous vehicles will benefit from this.”
Senior investor Eli Nir stated that Hailo has a world-class R&D team and innovative designs, making its chips the best partner for machine learning processing in edge computing devices.
“We are very pleased to provide Hailo with Series A funding,” Eli Nir said. “We believe that Hailo is a leading player in the search for the best AI/machine learning processing solutions. Its innovative solutions in edge computing can carve out a niche in multiple industries.【End】
Follow us on Sina Weibo, NetEase, Today’s Headlines, Daily Quick Report, UC Headlines, Sohu, and Yidian Zixun @AI-Drive.
Recommended Reading:
With Silicon Photonics Technology, FMCW Solid-State LiDAR System, This “Black Tech” Company Enters the AI Chip Market
In May, Turing Award winner and OURS CEO Tan Zhangxi, former Baidu Vice President and Kuandeng Technology Founder and CEO Liu Jun, Yu Wei Optical Chief Technology Expert Zhang Shi, and Yin Bing Technology Founder and General Manager Jiang Bo shared valuable industry experience from the perspectives of silicon photonic chips, mass-producible LiDAR mapping positioning and other modules, and achieving mass production of solid-state LiDAR chips. Click on the image to view last week’s member highlights.
In June, we will launch the “Smart Driving – Vision Special” where four mysterious guests will share their insights. Stay tuned.
