Meet Startup @TW

Taiwan’s Automotive Research & Testing Center to Build Autonomous Vehicles Locally

Taiwan’s Automotive Research & Testing Center (ARTC) will seek to leverage the country’s engineering expertise and global supply chain experience to build fully autonomous minibuses.

Beginning next year, the Chunghua-based government-funded non-profit research organization will establish a new subsidiary to oversee development of the vehicles, as Taiwan pushes to compete in the most dynamic and competitive space in the global auto market.

Residents of Taiwan have so far had little exposure to self-driving vehicles, save for a widely trumpeted test of a 12-passenger EZ10 bus in cities and university campuses around the country.

In May, French company EasyMile, debuted the EZ10 autonomous shuttle at Taipei’s National Taiwan University. The following month, the driverless shuttle made its way to the southern city of Kaohsiung, before moving on to Taichung, central Taiwan, and finally running a late-night test through the streets of Taipei.

Now Taiwan is ready to take its own self-driving vehicles off the test tracks and onto the streets.

Where Research Meets the Road

**The ARTC’s general road vehicle boasts several Advanced Driver Assistance Systems (ADAS) technologies such as image recognition, obstacle detection, lane assistance, self-parking, emergency braking, lane changing, automatic vehicle distance control, and many other autonomous functions that you would expect from a self-driving car. **

Caption: The App that allows passengers to call a car and take them to their destination.
吳晴中

The center has made great strides in vehicle convoy and lane changing technology, and have developed an app that allows passengers to call a car and take them to their destination – much like you would do with Uber.

Deputy general manager of the ARTC Ching-Chiu Liao was quick to point out a number of technical advantages the center’s model has over its driverless competitors. For example, EasyMile’s autonomous minibus relies on high-density maps and Light Detection and Ranging (LiDAR), a remote sensing method that uses a pulse laser to measure distances between objects that is often seen rotating on the top of many self-driving cars.

While this allows for accurate and automatic stopping and driving, the technology is currently unable to recognize traffic lights and therefore must rely on wireless data signals to calculate distances between vehicles and traffic signals.

However, ARTC’s technology utilizes a combined lens and LiDAR core which provides better usability in identifying lane markings, traffic lights, and pedestrians, according to Liao. "Our perception system is more diverse, it can follow the lane on the road in complex situations allowing the car to stop and go, and even change lanes, without the need for transportation infrastructure.We believe this makes our product far more competitive."

The Open Road Ahead

After three years of research and development, ARTC considers their self-driving tech to be on par with those trialed by overseas companies, and believes the time is right to pursue the next step. According to Liao, that will require locally manufactured cars. "Without your own vehicles, the situation is a lot worse. I really envy these French companies, they have their own funds to build cars.”

In order to reverse the situation and compete, ARTC intends to establish a new start-up company next year that will integrate Taiwanese component makers with an eye to producing a street-ready concept car in 2019. The center believes Taiwan’s experience supplying global electronics and vehicle brands with high quality components could provide a crucial cost advantage as the new car can be made entirely locally.

If Taiwan can develop autonomous vehicles, then there will be no need to spend extra time adapting and transforming foreign-built cars to meet its needs. "It's like a duck paddling in the water, slow is not necessarily bad. If you can use half the price to provide the same or even better functionality, then you will not necessarily lose," Liao said.

Liao concedes that without major car manufacturing in Taiwan, it will be difficult to compete, but made it clear that Taiwanese firms have a track record of innovation in autos that should stand the country in good stead. “If we start by creating a layout for the future now, we can train experts here and build the camera and radar technology that we need ourselves.“

Nearing the Finish Line

Liao stressed that as a nonprofit foundation ATRC must find a way to balance industrial development and technological breakthroughs at the same time, and is unable to operate like a major manufacturer. Instead, they must share responsibilities in order to focus.

For example, by using the Nvidia vehicle platform, ATRC is able to use pre-built hardware instead of diverting time and resource to developing a parallel solution.

Caption:ATRC is able to use pre-built hardware instead of diverting time and resource to developing a parallel solution.
ATRC

The cost is still quite high for all of the installed technology, running at over U$10,000 for the computers and laptops necessary to run the system. However, the team has developed a set of driving simulation programs that can train algorithms and accelerate the pace of the system’s machine learning.

The next milestone will involve presenting a modified car for use on the open road before the end of next year. Liao relayed that ATRC expects to operate outside of the Changhua Coastal Industrial Park with a test vehicle very soon.

During that process, the team will pay particular attention to safety and performance. If the test and the plans to bring local component partners on board are successful, Taiwan will move one step closer to manufacturing its own autonomous driving solution, and perhaps making an impact on the global market.

Original News is from Business Next.