2 min read
Obstacles Lay Ahead for Autonomous Driving
Driven by the IoT boom, cars are connected like never before. The connected car panel at CES Asia last week brought together prominent figures to butt heads over tone of the most interesting topics in the field: autonomous driving.
HD Maps & Autonomous Driving
High-definition mapping is the core of tomorrow’s self-driving cars. While regular maps are compiled for humans who can make their own judgments under certain conditions, HD maps are designed for machines, said Zhai Yao, a head of product development at Continental Automotive’s smart transportation unit.
“HD maps provide much more detailed information on road conditions which mainly fall into two categories: lane attributes, like width of roads, road curves, and object info, such as road signs and sentry box,” added Jing Muhan, the Vice President of mapping company NavInfo. It’s also about precision. A regular map can position a car within a meter, while HD maps can do it to within as little as ten centimeters.
The complexity of HD maps makes them more demanding to produce and upgrade. Zhai pointed out that crowdfunding maps is a possible solution. “Every vehicle can provide data to the backend to create a road model that can be used by all drivers.”
Kong Xianglai, the manager of Sogou Map, chimed in to add that it’s impossible to maintain a high mapping service quality if companies stick to a traditional industry chain division and rely on only a few mapping companies for mapping data supply. That holds especially true in China, where urban construction has brought constant changes to road conditions.
“We want to engage more end-users as mapping data contributors,” he said. He also noted that while it’s okay to involve individual users for updating POIs (point of interest), updating mapping data has a higher entry barrier in terms of technical concerns.
Jing sees the problem from a mapping service’s point of view. “HD mapping is still a highly professional field which requires professional technologies and data collecting equipment,” he says. More importantly, the government has control over mapping services due to security concerns. Also, it’s difficult for crowdfunding maps to fit in a market that has multiple map formats, said Jing.
Assisted Driving vs. Autonomous Driving
Internet companies and automakers take different views on boosting the market. Carmakers focus more on the vehicle itself with development for better radar or visual cameras, while internet companies want to push and share cloud-based information to cars, Zhai noted.
Their views towards autonomous driving are also divergent. Carmakers prefer to take a more conservative view in adopting smart transportation technologies. Assisted driving technologies, like self-parking and lane keeping are leaping into the market to boost sales, but automakers are very careful when it comes to fully autonomous driving technologies.
“The reason behind this is very simple. If self-driving vehicles become mainstream, cars will become a transportation device. When all the social attributes of car brands disappeared, people’s demand for purchasing their own cars will also decrease,”said Kong.
Meanwhile, internet companies led by Google lean towards fully autonomous driving experiences. “Internet companies enter automobile market as a disruptor,” said Kong.