Autonomous Driving and ADAS

Development of Autonomous Driving

The structure of the autonomous driving car , the type of the autonomous driving vehicle , necessary sensor and elements for Autonomous Driving technology is discussed here.

1. Mechanism of Autonomous Driving

In order to realize autonomous operation, it is the basic mechanism of autonomous operation to perform autonomous movement of the machine by linking three processes of "recognition", "judgment", and "operation".

1-1 Cognition

"Cognition" is to detect surrounding situations using sensors and others.There are various kinds of sensors to be used for autonomous driving. 3D-LiDAR which measures distance by skipping light, stereo camera which measures distance,a monocular camera for detecting an object. Various sensors such as GPS for specifying position information are used.We will recognize the surrounding situation from the measurement results of these multiple sensors.

1-2 Judgement

"Judgment" is to make a decision as to how to move a car (automobile) or a robot as a result of sensing. In this judgment the sensing result is input, programming which performs processing on that value, computation using Deep Learning (deep learning), etc., we will do the circumstances around the car and route selection (route plan) etc.

1-3 Operation

"Operation" is to do driving operation of a car or a robot on a computer based on the result of recognition and judgment. At automobile manufacturers, we control various types of actuators and electric signals from the installed ECUs and control the vehicles. ZMP utilizes the RoboCar series of robot cars to enable steering, brake and accelerator operation from dedicated ECUs. In particular, security measures such as responsiveness, reliability, and external hacking and takeovers are also important factors for machines in which people appear.


2. Vehicles subject to Autonomous Driving

Autonomous Driving technology can be applied to various industries. Basically it can be said that any machine that people drive will be subject to Autonomous Driving .

There is an airplane as a vehicle that has already adopted Autonomous Driving . In addition, all passenger cars equipped with Nissan's Serena etc, which is equipped with some Autonomous Driving (lane keeping assistance, front vehicle following, auto cruise control for all vehicle speeds) will be Autonomous Driving . Automation is also required in Japan where labor force is decreasing, such as agricultural machinery and construction machinery.

Also, in various countries around the world, various small mobility is developed as a delivery robot, and a service that delivers on-demand to the house in front of the house may be developed, as well as drone and home delivery service robot that was Autonomous Driving .


3. Sensor required for Autonomous Driving

There are various sensors required for Autonomous Driving, but here I would like to introduce about the sensors necessary for Autonomous Driving of the car.

3-1 Sensor that grasps the position of the vehicle

Self-location estimation (control to estimate your own position) is important for automobile Autonomous Driving .

There are cameras and 3D-LiDAR that can acquire the distance and shape of the surroundings as a sensor to be used for that purpose.
In addition, GPS sensor may be used in the meaning of acquiring position information.
For this GPS sensor, sensors for outdoor use are common, but localization using GPS sensors is also possible indoors.
The GPS acquires multiple signals from the satellite and calculates the position, but the accuracy of the position information deteriorates in places where the satellite does not reach (inside the building or shade of the high building). Therefore, in indoor, a device that transmits signal information (latitude and longitude) of position information is installed on the ceiling, and a method of grasping indoor position is also used for indoor automation.

3-2 Sensor to monitor the surroundings

After an automobile or machine grasps its own position, what is around your vehicle is important to move autonomously. There are various sensors to monitor the surroundings, but below are representative sensors.
9. 3D-LiDAR
Currently, 3D-LiDAR can be cited as a sensor that is used in Autonomous Driving vehicles. This sensor is a sensor to measure the distance by skipping light. Although there is a difference in distance, if it says in person, is it an image that grasps the shapes and positions of surrounding objects and sees the surroundings by hand and moves it?
At the same time as this sensor is utilized in the above self location estimation, it judges the object recognition and the presence or absence of the object which estimated the shape from the information of the acquired point cloud.
The camera is applied to distance and object recognition using the image information. Because it hits the part of "eyes" that people say, I think that it is a sensor that is easy to image. The camera also has a monocular camera and a stereo camera.
Monocular cameras are used in the form of objects, white lines, and signal recognition. The stereo camera calculates the distance of the object ahead from the difference (disparity information) between the two cameras.
· Millimeter wave sensor
The millimeter wave sensor which is still mounted still transmits ultrasonic waves and detects the existence of objects ahead from the interference. It seems that it is possible to perform tracking by attaching the ID of the object with the algorithm inside the sensor.
Since the output of the car-mounted millimeter wave sensor is made with CAN, the data is light and easy to handle.
Millimeter wave sensor is not affected by bad visibility, so you can detect things even in bad weather such as rain, snow, dense fog.
· Sonar sensor
The sonar sensor which is adopted as a proximity sensor of a vehicle (sound when going closer, etc.) is also utilized as an auxiliary sensor for Autonomous Driving . America's Tesla is a concept to implement Autonomous Driving a camera and sonar sensor configuration, it seems possible to distinguish distant recognition using a camera and using a sonar sensor when approaching. Millimeter wave sensors and sonar sensors have relatively small data volumes, so they have features such as high speed computation because the load at calculation is small.
There are various sensors for monitoring the surroundings, but not all of them can be grasped by any one, but it is required to integrate multiple sensors and to reliably grasp the surrounding environment.
To that end, the existence of the system to be used for Autonomous Driving described next becomes important.


4. System used for Autonomous Driving

Autonomous Driving , it is necessary to handle the sensing result correctly, make a judgment for the next action, and do the operation by attaching the sensor, not the end. The one that realizes the series of operations becomes the system of Autonomous Driving .

In the automobile industry, there are cases in which each automobile manufacturer independently develops it, and in some cases it is commonly released as an Autonomous Driving system and provided as a tool to accelerate development competition.

In ZMP, IZAC (Isaac) has been developed as a system of Autonomous Driving control, and famous as a system that is publicly publicly available is Autoware (Auto) as an open source of NVIDIA's Autonomous Driving platform and Autonomous Driving system Wear) and Appolo (Apollo).

5.Autonomous Driving Car traveling area

Autonomous Driving is realized, I think that it is desirable to apply it in various places, but I think that the technique of Autonomous Driving will be used gradually.
Therefore, it is thought that the running area of ​​the Autonomous Driving car will also gradually expand.

Clear criteria are not decided, but car makers conduct full Autonomous Driving on public roads in late 2020 or 2030. In the meantime, it seems that partial Autonomous Driving will be done, such as Autonomous Driving on the expressway and automatic parking at the parking lot.

Also, the technology of Autonomous Driving will be more prevalent in closed places such as private estates and test courses than vehicles running on public roads.

Even with ZMP, we are working on automation of tire evaluation of test course through cooperative development with Bridgestone and automation of construction machine with KOMATSU in construction machine industry.

In closed areas such as test courses, it is said that there are environments where Autonomous Driving techniques are easy to put into practical use, as hurdles such as the ability to control external environments to some extent and legal issues are low.

As for the Autonomous Driving of public roads, it seems that it is necessary in the future to aim for practical application by proceeding with technical accumulation through demonstration experiments and service verification, improvement of social acceptability of users, and improvement of regulations.

6. Autonomous Driving development solution

At ZMP, we are working on the development of Autonomous Driving vehicles aiming at practical application of Auto taxi in 2020. In August 2018, we carried out the world's first Auto taxi service trial and are working on the development for service commercialization.

ZMP provides an Autonomous Driving system combining mobility and service, MaaS (Mobility as a Service) to service providers who want to utilize Autonomous Driving for service development and verification.

We are actively working on automating all moving bodies regardless of passenger cars. In order to realize the Autonomous Driving service created by the service provider we are offering the system related to Autonomous Driving which we have developed so far as an "Autonomous Driving platform" to business operators aiming to commercialize MaaS.

For details please see the following page.

7. Inquiry/Document request

For inquiries regarding the above & document request etc, please contact us.