Baidu could seek sensor system targets to grow intelligent vehicle business division – sources
15 JUN 2017
Baidu [NASDAQ:BIDU], the China-based search engine, could seek sensor system targets to grow its intelligent vehicle business division, according to two sources familiar with the situation.
The company would most likely scout for such targets in Israel, the US and Europe, where sensor technologies applied in autonomous driving system are better developed and studied on, the sources said.
It welcomes advisory approaches regarding target referrals, and would closely evaluate if potential targets are in accordance with the general strategies of its intelligent vehicle unit, as well as to see if targets could fit in and add on to Baidu’s current technology in autonomous driving, the second source familiar said.
The company would not place a range on the size and earnings of the potential targets, both sources noted.
The USD 61.58bn market-cap tech giant is currently trading at a P/E ratio of 46.4x. It reported annual revenue of CNY 70.5bn in 2016, up from CNY 66.4bn from the previous year, according to its annual report. Net income declined from CNY 33.3bn to CNY 11bn in 2016, the same report shows.
The company had cash and equivalents of CNY 10.9bn as of the end of 2016, the annual report read.
Baidu is working to offer technologically mature sensors at a reasonable price, which would be the key in developing autonomous driving in the real world, said Weihao Gu, General Manager of Baidu Intelligent Vehicle Business Unit during the 2017 Commercial Electronics Shows (CES Asia) held last week in Shanghai.
The company started investing heavily on the research and development (R&D) of autonomous driving technologies since 2015, as reported. It performed self-driving tests on Beijing’s highway and city roads in the end of 2015, gained self-driving road testing license in California, the US, in September 2016.
Also in November last year, Baidu launched a trial operation of its autonomous driving vehicles on the roads of Wuzhen, of China’s Zhejiang province.
The company relies on deep learning to transform data collected via man driving on different road conditions, under various weather conditions, into behavioral knowledge, which would ‘command’ the self-driving system to make similar decisions as mankind when faced with similar road or weather conditions, he explained.
Such transformation relies on efficient algorithm to ensure the speed and accuracy of deciphering big-volume data. Baidu relies on Convolutional Neural Network (CNN) to automatically collect and identify multi-layer images, which is used to control steering, or the horizontal movements of the self-driving vehicle, Gu noted.
The company also introduced long short-term memory (LSTM), a recurrent neural network (RNN) architecture, which is able to make predictions on ‘recurrent’ behaviors. This algorithm is very effective in improving the vertical control of the self-driving cars because it contains ‘time memory’, namely it tries to imitate humans in their choice of speeding up and slowing down the cars under different conditions, which would smooth the vehicles’ self-control, he explained.
Baidu is intending to open up its self-driving miles data to the wider developer community via Baidu Computing Unit (BCU) platform, Gu said. This will also be part of the company’s Apollo project, which is to be launched in July 5th in Beijing.
Apollo is expected to lower the threshold of autonomous driving industry, as it contains the architectures of Baidu’s self-driving technologies, which would potentially allow any capable developer to manufacture a self-driving car based on the existing data of environment perception, itinerary planning and vehicle control, among others, he said.
Baidu has formed strategic alliance with Bosch and Continental [FRA:CON] earlier this month, to cooperate on the fields including autonomous driving, intelligent transportation and internet of vehicles, as reported. The company, together with Ford Motor Company [NYSE:F], invested USD 150m in Velodyne LiDAR, a recognized global leader in Light, Detection and Ranging (LiDAR technology), as announced.
It also participated in the latest fundraising of NextEV [Wei Lai Qi Che], a China-based, privately held intelligent electric vehicle R&D company in March, as reported. NextEV was said to be valued at CNY 20bn upon completion of the financing.
Baidu declined to comment.