With cognition, you are required to know the circumstances around your car.
In a car, the driver
· Environment around your vehicle (cars, people, signs, etc.)
· Position where your vehicle is located (driving place, driving area (direct route, intersection etc.))
· Behaviors of the vehicle (accelerating, decelerating, etc.)
As you know, you can decide the next action (acceleration, deceleration, turn).
The driver can grasp the current situation from his or her eyes and direction feeling, information on the land intuition and the route up to that point, the stepping angle of the accelerator / brake, and the angle of the steering wheel.
What do you do with Autonomous Driving so that an Autonomous Driving system replacing a person.
[Know the surroundings of the vehicle]
Autonomous Driving vehicles, we have a camera that plays the role of human eyes and LiDAR that measures the distance by hitting light to gather information on the surroundings. With this camera, cars and people, signs, signals, white lines etc. are recognized. In addition, LiDAR measures the distance three-dimensionally depending on the type and measures the surrounding shape and the distance to surrounding objects.
There are two types of cameras, one monocular type and one stereo type used with one camera. Even though you think that you also have experience, just like stereotyping is easier to recognize the distance to the object than the monocular type, as well as the sense of distance becoming difficult to grasp one eye, the point of measuring the distance in terms of performance it is excellent.
However, because high-performance processing is becoming possible even for cost, installation location, monocular type, depending on the type you choose to realize.
For example, if you go for examples of vehicles that are on the market today, Nissan Serena 's "Pro Pilot" and Toyota' s safety sense etc. adopt monocular type.
Also, cameras are the same as human eyes and the detection accuracy will degrade at nighttime or in bad weather like rain or fog.
Therefore, for the purpose of assisting the camera, sensor fusion technology using infrared laser and millimeter wave radar is important, which can detect the distance to the surrounding objects in a wide range.
Even in Tesla's vehicle that implements the system of Autonomous Driving, the main awareness of the environment around the camera using the camera is also carried out, but also equipped with radar and ultrasonic sensor to assist it.
For cameras, image processing is more important than acquiring images. Various image processing such as whether you want to detect distances and obstacles like stereo cameras or whether you want to recognize from images is currently being developed.
[Know the position where your car is located]
For confirming the position of the vehicle, GPS which has already been put into practical use in car navigation systems etc. is mainly active. There are several kinds of GPS, and if you want to use it as a control of a vehicle of Autonomous Driving , accurate position information (errors in centimeters) is required.
For Autonomous Driving public roads, maps are important to keep track of where your car is now. Regarding maps that are called high precision maps with high accuracy and high information volume are required. This high-precision map is created based on map creation equipment developed by Mitsubishi Electric, such as MMS (mobile mapping system), and ZMP develops and creates technology to create necessary maps for its own Autonomous Driving.
Moreover, in order to accurately grasp where the vehicle is in the utilization of this high-accuracy map, we used point cloud information acquired by 3D-LiDAR and white line detection of the camera.
[Know the state of the car]
Autonomous Driving , is my car currently accelerating? Or is it slowing down? Or is it curving? You need to know about.
Therefore, in order to grasp the state of the own vehicle, a 6-axis inertial sensor which can detect acceleration, angular velocity (roll, pitch, yaw) of X, Y and Z axes is installed.
Also, at the same time, from the driving information of the vehicle, the movement of the own vehicle is judged and more accurate control is carried out.