Autonomous Driving Technology: Telematics Japan 2014

Results of Mercedes-Benz tests; OEM commercialization timings

2014/11/11

Summary

 This report gives a synopsis on some of the lectures that were given on autonomous driving at Telematics Japan 2014, which was held at the Hilton Tokyo on October 15th and 16th and sponsored by Telematics Update.

Included are:

* The roadmap for autonomous driving by Mercedes Benz, Nissan, and Honda, which envisions the reality of autonomous highway driving by 2020.

* Issues that Mercedes Benz faces in perfecting autonomous driving. The automaker's autonomous driving test conducted in 2013 proves that a high level of autonomous driving is possible, even under complex traffic conditions such as city driving.

* Initiatives for creating highly accurate map data needed for autonomous driving, presented by HERE, a European company that provides maps through cloud technology.

* Initiatives for providing information on autonomous driving based on using telematics, which includes a huge assortment of information; and for standardizing information globally.


Related Reports:
ITS World Congress 2014 (in Detroit) :

CTO plenary session overview
Exhibits and demonstrations


ITS World Congress 2013 (in Tokyo) :

Autonomous driving technology beyond reality?
Toyota & Honda demonstrate autonomous driving
"Connected vehicle" becomes a new form of safe driving



Roadmap for autonomous driving

 Mercedes Benz, Nissan, and Honda all made presentations on their roadmaps for making autonomous driving a reality. Based on the precondition that drivers will at all times be able to take over the control of their vehicles at any time, the companies outlined the details of how they plan to make autonomous highway driving a reality by 2020.

Mercedes Benz


Speaker:
Dr. Hartmut Schaefer
By 2020
Phase 1: autonomous highway driving in traffic jams (at slow speeds)
Phase 2: autonomous parking
By 2025
Phase 3: autonomous highway driving at high speeds
Phase 4: autonomous driving under conditions other than high-speed highway driving
Nissan


Speaker:
Koji Yamamoto
By the end 2016
Launching the "traffic jam pilot", which is technology enabling autonomous driving on congested highways
Launching autonomous driving technology that eliminates the need for driver input
By 2018
Launching multi-lane autonomous driving technology that avoids dangerous situations on the highway, and that enables automatic lane changing
By 2020
Launching autonomous driving technology that makes it possible for vehicles to go through intersections and crossroads, without the need for driver input.
Honda


Speaker:
Toshio Yokoyama
By 2016
Enabling continuous driving on the same lane (driving assistance)
Between 2020-2021
Enabling continuous driving on the main lanes on highways (driving assitance)
By 2030
Enabling optimum driving on all congested highways, including conecting roads. (Expanding driver assistance: drivers monitor systems during ordinary driving conditions, but need to control vehicles in emergency situations)

 

 



Autonomous driving tests by Mercedes Benz

testing actual automated-driving technology Ahead of other OEMs, Mercedes Benz demonstrated the feasibility of autonomous driving technology for 100km on public roads, including those in cities.


 Dr. Hartmut Schaefer, Senior Expert, Mercedes-Benz R&D Kawasaki, Mercedes-Benz Japan Co., Ltd. gave a presentation on the results of experiments that Mercedes Benz's performed in 2013. In this report, Marklines has compiled Mercedes Benz's development policy of autonomous driving, details of test results, including successful outcomes and specific issues that need to be addressed. The summary is based on information that was presented at Telematics Japan 2014 and on the information contained in Daimler's press releases announced in September 2013.


Source: Daimler

Summary of Mercedes Benz autonomous driving test

Objectives 1. Making "accident-free driving" a reality is one important step in this direction.
2. Increasing driving safety by finding solutions to driver inattention and error.
3. Freeing drivers from long, boring drives and difficult driving conditions.
Test details The test drive was not conducted on specially built testing roads but on actual public roads in cities and under everyday traffic conditions. In other words, the test was done under everyday driving conditions in the presence of unpredictable obstacles such as pedestrians, bicycles, parked cars. In addition, no lead car was used. The company proved that a high level of autonomous driving is indeed possible, and gained insights about specific issues and the direction in which the company needs to further develop its current systems.
Test route The route was from Mannheim to Pforzheim, a distance of 100km. It required the vehicle to outmaneuver rather intricate and difficult driving conditions that included heavy traffic, signals at intersections, pedestrians, bicycles, streetcars, and other complex situations. This route was chosen to commemorate the first long-distance route traveled in the world by Mercedes Benz's founder's wife, Bertha Benz,125 years ago for the first time in the world. The test car's following this same route added historical significance to the test.

Test route

Test route
Source: Daimler

 

Research vehicle

A new Mercedes-Benz S-Class was used. The company did not use any special sensors for the test. Instead, the company relied on basically similar sensors and cameras that are currently available on the mass-marketed S-Class. By increasing the number of sensors and cameras, the final operating system was capable of seeing a full, 360-degree view. Due to this, the system can be easily mass produced.

Test car

Test car
Source: Daimler

 

Research-car system

Sensors and cameras The company used sensors and cameras similar to those equipped on its current vehicles, making it easier to easily equip existing technology to production vehicles in the near future.

Sensors and cameras

Source: Daimler

Long-distance radar sensors
Two long-distance radar sensors were added to the left and right sides of the front bumper, making it much faster for the vehicle to detect vehicles approaching from the left and right directions at intersections. In order for the vehicle to detect road conditions behind it, the company installed one long-distance radar sensor at the vehicle's rear.
Close-distance radar sensors
Close-distance radar sensors were placed in all four corners of the vehicle, increasing the vehicle's ability to detect nearby objects, pedestrians, bicycles, etc.
Stereo cameras
The company improved the accuracy of the vehicle's ability to measure distances, by expanding the distance between the camera lenses. Objects that were in locations further away were able to be detected not only by radar but also by cameras.
Wide-angle cameras
The company incorporated a 90-degree wide-angle lens in the windshield to enable the vehicle to detect signals at intersections.
Rear-view camera
One rear-view camera was installed at the rear of the vehicle to enable the vehicle to see the rear area. By comparing stored, 3D-digital-map data showing the features of surrounding areas, it became possible to better identify the vehicle's location with higher accuracy than by using only conventional GPS.
Control system The vehicle's location could be detected from information obtained from sensors and cameras. The system compares the information with digital maps that had been created especially for the test. The system further analyzes routes that the vehicle could safely travel through, making it possible for the vehicle to set its course. The algorithm needed to accomplish this was developed by Mercedes Benz's research team jointly with the Institute for Measuring and Control Technology at The Karlsruhe Institute of Technology. With the aid of its highly automated "Route Pilot", the vehicle was able to determine its appropriate route through traffic, facing a variety of scenarios. For example, oncoming traffic at rotary intersections and crowded areas, bicycles on the road, cars parked in all sorts of ways, red stoplights, intersections with through streets, pedestrians crossing roads, and streetcars.
Digital maps Based on cooperation from Nokia and HERE, Mercedes Benz created 3D digital maps showing the route between Mannheim and Pforzheim, which were explicitly needed by the research car. These maps required an extremely high level of precision and accuracy, such as the exact number, location, and direction of lanes, traffic signs, and signals.

 

Achievements of the test

Under the precondition that drivers must constantly keep their eyes on the road at all times and have the ability to take control of the vehicle in emergency situations, Mercedes Benz proved that highly accurate autonomous driving can be achieved both under highways and simple traffic situations. In addition, the greatest achievement that resulted from conducting an autonomous driving test on public roads was that the research team was able to discover what areas they need to focus their attention on next and think about which direction they should advance their research further. They left a record describing every type of situation in which driver input was needed, and by reviewing the record and conducting evaluations accordingly, they are now in a position to expand the range of autonomous driving operations. This will result in increasing R&D achievements and expanding the scope of drivable road conditions.

 

Issues identified

Identifying the colors of traffic signals The research car had difficulty detecting the colors of traffic signal lights, faced with so many lighting conditions. It was also difficult for the vehicle to accurately determine which traffic signals corresponded to which driving lanes. If a solution to these issues cannot be found, autonomous driving in city areas will be impossible.
Driving operations Instructions controlling the steering, engine, and braking were programmed to respond to various situations such as the vehicle's passing through roundabouts. The research team identified which operations need further improvement and refinement.
Determining positioning/ mapping Further accuracy is needed to fix the vehicle's position or placement on the road. For example, when stopping at intersections, the vehicle needs to determine exactly where it needs to stop. Since the vehicle needs to check for other traffic going through the intersection, it has to search for the right place to stop, making the actual positioning of the vehicle on the road extremely important.
Social issues One of the big issues when driving is communicating and interacting with other drivers on the road. If there are obstacles in the road, drivers need to carefully scrutinize the situation and decide which vehicle will go by the obstacle first. Human drivers will be more bold in deciding that they can go ahead, but autonomous driving controls with a higher degree of caution.
For example, sometimes pedestrians wave to vehicles stopped at crosswalks in order to signal to them to go first. But it became evident from the test that the research vehicle had difficulty determining what pedestrians are doing or are going to do. This is because, at the time the vehicle was programmed, no one had given a thought to common road manners and courtesies exchanged among drivers and pedestrians. Nevertheless, the researchers aren't thinking of responding to each of these types of situations. For example, when garbage trucks block the road, it is not a good idea for self-driving vehicles to automatically pass them, since the sensor's field of vision ahead is limited. This is one type of situation requiring driver input.

 

Acceptance by society in general

There is one requirement that must be met so that society will accept autonomous driving, and the requirement is the exact, same one that was needed when automobiles themselves were first launched. A certain amount of time needs to pass before society gradually gets used to the idea that autonomous driving is reliable technology. In a research study conducted by Mercedes Benz, 100 individuals between the ages of 18-60 were invited to experience autonomous driving for themselves in a simulator. The company found that practically all of the participants' initial skepticism disappeared after they had actually experienced autonomous driving. Even participants who had extremely negative opinions from the start ended up very receptive to autonomous driving after trying it.

 

Practical utilization of telematics

Car-to-X Communication is being proposed as one of the ways to ensure that maps and route information always contain the latest information. If this communications technology is put to practical use, it will be possible to generate maps in real time based on vehicles communicating among themselves. In other words, Car-to-X Communication is designed to enable vehicles to record their routes and save that information in databases. For example, if one car stops at a red light, that vehicle can communicate that it is stopping, to other vehicles on the road. Until now, information drivers obtained was limited to only what they could see directly ahead of them, but with this latest telematics technology, drivers will be able to receive information about what lies further ahead on the road. For many years, Mercedes Benz has been advancing technology on vehicle-to-vehicle (V2V) communication and vehicle-to-infrastructure communication. Now, the company plans to become the first OEM in the world to launch series production vehicles equipped with Car-to-X technology.

 

 



Map data and telematics required for autonomous driving

 Marklines compiled the following information on map data and telematics for autonomous driving based on the speech presented by Mr. Floris van de Liashorst of HERE, a European company providing maps through a cloud server.

3D map data required for autonomous driving

The level of accuracy needed for pinpointing vehicles' precise locations during autonomous driving is much greater than that required by conventional car-navigation systems that rely on GPS. In addition to two-dimensional map data, autonomous driving requires 3D map data, which adds a new dimension in terms of not only showing road layouts in more detail but also giving precise information on the locations of lanes, stop lines, guardrails, and on the shapes of buildings. In order to obey traffic rules, self-driving cars need to know the exact locations, numbers, and directions of traffic signals and traffic signs. Furthermore, it is extremely important that information about any changes in road conditions, such as the narrowing of roads due to construction, be instantaneously updated in real time.

 

Role of cloud servers

Role of cloud serversHigh precision maps require an enormous amount of information. Therefore, it seems to be an appropriate solution to store information on the cloud and transmitting it back and forth between vehicles traveling on the road. Cloud servers can provide information to self-driving cars about traffic conditions, accidents, obstacles on roads, in addition to street maps. If more and more vehicles are enabled to communicate via telematics, every car on the road can become sensors gathering information on the latest road conditions and providing it in real time to other vehicles traveling in the area.

Source: HERE

 

Autonomous driving will never be possible without telematics

Thanks to telematics, maps will become even more sophisticated as real-time data on driving conditions are stored in cloud servers. Whenever a change occurs, the cloud can transmit that information instantly to vehicles. Furthermore, significantly more valuable information could be provided to self-driving vehicles if more precise data could be transmitted to them, including information on the proper ways they should handle road everywhere, such as how to turn at certain intersections and how to reduce speed when exiting certain freeway off-ramps. In other words, telematics plays a key role in making autonomous driving a reality through the transmittal of information. Recently, international discussions have begun on how map data and communication systems should be standardized, considering the huge variety of information available.

<Automotive Industry Portal MarkLines>