Naoki Suganuma, the head of Kanazawa University’s Autonomous Vehicle Research Unit, gave a lecture in Tokyo on Sept. 15.
The Ph.D. engineer delivered a presentation titled, “The future of autonomous vehicles,” to an audience that included many people from the automobile industry. The lecture, organized by The Japan Times, took place at Nifco Hall in Tokyo’s Minato Ward.
After a greeting from Japan Times President Takeharu Tsutsumi, Suganuma began his presentation by introducing the members of his Autonomous Vehicle Research Unit — seven teachers and researchers, and 11 students.
The members are not only technical experts, but include one specializing in economics because the economic aspect will be important when autonomous vehicles become a reality, Suganuma said, adding that he likes the size of the unit, 18, because it makes the decision-making process so much quicker.
He started by explaining the history of autonomous vehicles, or as they are sometimes called, driverless or self-driving cars.
“Autonomous driving is actually not a recent development. In Japan, a group of researchers conducted test-drives of an autonomous vehicle in the 1960s,” he said.
Autonomous driving is sometimes a government project as countries aim to support companies at home to develop the key technology. In Japan, the Ministry of Economy, Trade and Industry led a project, dubbed “Energy-ITS (Intelligent Transportation System),” from which Suganuma played a video of three trucks running exactly 4 meters apart as an example.
The project was conducted from 2008 to 2012 as part of government initiatives to realize Japan’s international pledge to cut carbon dioxide emissions. Energy-ITS was found to cut the emissions by 18 percent, he said.
In the early stages of autonomous driving development, demonstrations were conducted only on expressways, but recently the enhanced performance of computers and precision of digital maps have enabled test-drives to be carried out on regular roads.
“Earlier, they relied a great deal on hardware infrastructure, such as burying magnets under paved roads. Nowadays, we increase our reliance on software infrastructure such as better road maps. That speeds up autonomous driving development because altering hardware infrastructure is too expensive,” he said, adding that the motivation of the development was to enhance the safety and efficiency of logistics, and handle rapid increases of traffic. It will also become an alternative transportation method in areas where the population decreases and the ratio of elderly people increases.
Suganuma highlighted examples of large-scale autonomous driving demonstrations that boosted the momentum of development. The Defense Advanced Research Projects Agency (DARPA) Urban Challenge was held Nov. 3, 2007, in California. The event, involving Stanford University, Carnegie Mellon University and other organizations, required teams to build an autonomous vehicle capable of driving in traffic, and performing complex maneuvers such as merging, passing, parking and negotiating intersections.
In Japan, Tokyo Motor Show 2011 had a demonstration of autonomous driving, provided by Toyota Motor Corp. and Kanazawa University. Nissan Motor Co. and the university held a demonstration at major trade shows such as ITSWC 2013, CEATEC 2013 and Tokyo Motor Show 2013.
“Since 2013, autonomous driving has been a very hot topic,” he said.
Suganuma then turned to his unit’s activities.
Kanazawa University began autonomous driving research around 1998.
“Back then, carmakers were developing autonomous driving on expressways. Kanazawa University wanted to do something different, and better, and thus aimed to test drive autonomous vehicles on regular roads,” he said.
In 2008, the university’s research team conducted unmanned autonomous driving on campus, he said, adding that only three people watched the demonstration.
“But that quickly changed when we successfully drove autonomous vehicles on public roads that are not expressways in 2015. It generated so much buzz that the whole of society was excited,” he said.
And much of that excitement is being created by the unit’s latest research car, the university’s fourth. The car, a Toyota Prius, comes with all the equipment necessary for the car to drive itself — such as antennas, sensors and laser-emitting devices.
He explained the car’s Laser Rangefinder, or LIDAR, technology and displayed the image LIDAR “sees,” showing how the car recognizes other cars, pedestrians and various objects. He also introduced another technology, the millimeter wave radar, which provides clear vision even in rain or snow.
“Those sensor technologies enable recognition of the environment surrounding the car in the complete dark, which means the ability exceeds that of human beings,” he said.
He introduced technologies to enhance one of the key aspects for autonomous driving — the ability of the car to work out its location. This is important in case, for example, tectonic plate movement alters the location of roads from the location on the digital map autonomous vehicles use. It is especially important in quake-prone Japan.
Other technologies Suganuma introduced were able to predict what is not visible but will potentially become a danger.
Then, Suganuma discussed demonstrations his unit conducted.
His team’s first test-drives were at a driving school in Kanazawa from January 2013 to February 2015. The tracks at driving schools tend to be difficult because they are designed for students to practice many driving techniques such as right turns and parallel parking.
“We rented the entire school track for our exclusive use every weekend. The practice at a driving school was definitely useful,” he said. It particularly helped develop so-called path planning — a function to get to destinations by following traffic rules and avoiding accidents.
Next, Suganuma discussed on-road demonstrations in Suzu, Ishikawa Prefecture, of which Kanazawa is the capital.
Suzu sits at the tip of Noto Peninsula. The small city’s population was 15,948 and the ratio of elderly people was very high, 44.2 percent, as of December 2014. Suzu City Hall is three hours from Kanazawa Station by highway bus. The city has no rail service but there are some buses and taxis.
These factors add up to concerns about transportation systems in the not too distant future, and autonomous vehicles may be able to contribute to solving such problems, Suganuma said.
Hence, Kanazawa University began autonomous driving tests on roads that are not expressways Feb. 24, 2015, making it the first Japanese university to do so.
“The fact that roads are narrow was not a major problem. However, a mountainous area where the GPS (global positioning system) signal was weak was a challenge,” Suganuma said as he described the first tests on a 6.6-km route incorporating urban and mountain driving, where some parts of a road on a digital map are 2 meters off the actual road.
His team stopped using GPS in the area where the signal was weak, which turned out to be the correct decision, he said. On April 1, 2015, a one-way test-drive was successfully completed and on April 16, the autonomous vehicle made a successful round trip.
The team continued the experiment on a second route, 60 km long, in Suzu and other routes in Kanazawa.
Suganuma played a video showing the vehicle stopping even at a green light when a bicycle crossed in front of it after ignoring a red light.
“Human beings get tired processing multiple contradicting information such as a bicycle ignoring a traffic light, green lights and traffic in the opposite lane. But the autonomous driving system never gets tired,” he said.
Suganuma went on to discuss the results of a questionnaire on autonomous driving. The poll involved 4,000 households in Suzu in October and November 2016, of which 631 responded. The respondents did not include minors.
To the question whether autonomous vehicles are necessary, 54.2 percent said yes, 18.7 percent said no and 27.1 percent said neither yes or no. On what level of autonomous driving is desirable, 40.4 percent wanted Level 4, in which drivers do not have to do anything, while 31.3 percent wanted Level 3, or would drive only in case of an emergency.
Level 2, a function to prevent a vehicle drifting out of a lane, got 15 percent, while Level 1, automatic braking, was sought by 12.9 percent of respondents.
While autonomous driving is something society is waiting for, some challenges remain and Suganuma listed them, including insufficient functionality of sensors, inability to deal with extreme weather such as heavy rain and snow, and the inability to recognize unusual gestures / movements by pedestrians, traffic controllers at construction sites and police officers.
The Geneva Convention on Road Traffic stipulates there must be a driver in a moving car and many countries have ratified it and made it illegal to have a car without a driver, and thus there must be legal changes to promote autonomous driving.
There was a Q&A session after Suganuma’s presentation.
Asked if communication technology can be a key technology, he said it can be when the autonomous driving system has to “see” what is not visible.
He added, though, that because there are so many cars, handling large amounts of communication is a challenge.
The second question was about car insurance for autonomous vehicles. He said insurance is part of his research and someone in the insurance industry had told him that normal insurance is good enough, he said.
“However, from now on, there will be no drivers and only passengers in autonomous vehicles. In that case, who would accident victims sue for damages? Responsibility may lie with the makers? Legal changes will have to be made,” he said.
To a question on whether Level 3 is a realistic choice, Suganuma said, “Personally, no.” His opinion is that practical use of autonomous driving will be divided into two patterns — 1) very convenient driving support, and 2) completely autonomous buses and other public vehicles.
The reason why he thinks autonomous passenger cars will not be widely used is that sensors are expensive and hard for individuals to maintain.
Another inquirer referred to the planned launch of satellite Michibiki in October and asked about the reliance on GPS.
Suganuma said GPS is not totally unnecessary for autonomous driving and he welcomes improvements in the system. GPS can be the basis of information on location, but autonomous driving researchers should not rely too much on it to create and update a map, he opined.
Asked if automobile makers are competing or collaborating on autonomous driving, Suganuma answered there will probably be competition on things that influence the comfort of passengers.
“However, for example, let’s say buses become autonomous vehicles. I can’t imagine the market will expand to the extent where economies of scale work. That would mean carmakers will want to collaborate rather than compete,” he said.
This page has been produced with the support of the Ogasawara Foundation for the Promotion of Science and Engineering, which was founded by Toshiaki Ogasawara, the former chairman and publisher of The Japan Times and the chairman of Nifco Inc.
Naoki Suganuma obtained a Ph.D. in engineering from Kanazawa University in 2002. After taking up a Research Fellowship for Young Scientists at the Japan Society for the Promotion of Science, he began working as an assistant at Kanazawa University’s College of Science and Engineering.
In 2015, he was transferred to the university’s Institute for Frontier Science Initiative, whose main focus is interdisciplinary research. He currently heads the Autonomous Vehicle Research Unit as well as being an associate professor.
He began research on autonomous vehicles in 1998 and experiments on driving the vehicles on public roads other than expressways in 2015, making Kanazawa University the first Japanese university to do so.
He won the Masao Horiba Award, which was created to encourage researchers and engineers involved in analysis and measurement technology, in 2016.