Lei Feng network press: author of this article Zhu Yulong, automotive industry practitioners, the main car electronics.
In simple terms, it is now a German car company that has taken US and global technology companies together to create a digital computing platform that will cooperate with Germany's chassis electronics and EE bus to innovate and evolve.
Recently, BMW, Intel, and Mobileye opened a small party and organized a group to compete in the autopilot war. We should also face this war in the auto industry for the next 100 years.
The three parties will jointly develop self-driving cars based on the BMW i Vision Future Interaction concept car
Production version will be launched in 2021
The three parties have created a better platform for self-driving.
We simply consider, especially taking AUDI to PK about BMW
1) ADAS field
In fact, we seriously compare, the two ADAS in L0 ~ L1 in fact the difference is not.
Sensor Configuration Find Bosch, Conti, Kostal, etc.
77G Radar will converge in addition to the price (you can choose the same)
ME+ camera module is the same
2) Automatic driving
In accordance with their two concepts, BMW and AUDI have the same ability to find sensors.
Both IBEO lidar will give
What difference do we consider? What is worth to draw?
This group of BMW
1) BMW takes the lead in coordinating the car
1.1 Fixing Radar&Lidar&Camera Sensors
1.2 Get the actuator
1.3 Adding V2X&Telematics
2) Intel provides computing platform
In May 2014, Intel contributed 100 million knives to the ZMP. ZMP's "IZA (Intel ZMP Autonomous Computer)" is equipped with real-time Linux box devices. The processor is equipped with Intel's "Core i7". The IZAC connected to sensors such as LIDAR, camera, millimeter-wave radar, and GPS is connected to the vehicle's CAN bus, and the autopilot application software runs on Linux.
3) ME provides deep learning around the image
This is actually a point in Amnon Shashu's opinion. I agree: the three pillars supporting autonomous driving
1. Sensing: This ME company provides visual extraction technology while BMW sets up a complete redundancy system.
The sensor combination senses the road environment and obtains the environmental data. Through the calculation unit, the environmental model is established.
2. Driving policy:
Autopilot system needs to learn how to abide by traffic rules => DNN and Rule based
When to go
When to stop, etc., all need
The system needs to organize these into rules that the system complies with
3. Mapping
The autopilot system needs to create a very accurate map to deal with road conditions
Also determine where you are
The redundancy system established by BMW here
Perceptual Redundancy: Sensor Fusion
Map & Positioning Redundancy => Path
High-precision map & REM redundancy
GPS+IMU positioning redundancy
Compared to AUDI, this pattern has taken a step forward.
Assist: Altera's Cyclone V SoC FPGAs: Processing Logic
redundancy
32-bit TriCore Fail Safe
Core processing section
NVIDIA Tegra K1 Processor: A Large Number of Operations
Periphery
Camera = "MobilEye EyeQ3: Handle the front camera, here is the supporting role.
Ultrasonic
77G Radar
24G radar
Ultrasonic
In a sense, the success or failure is determined by the full use of the visual system, which determines the cost and efficiency of the entire system.
In contrast, I believe that AUDI also needs to pull more helpers to fight together. In simple terms, it is now a German car company that has taken US and global technology companies together to create a digital computing platform that will cooperate with Germany's chassis electronics and EE bus to innovate and evolve.
summary:1) Tesla's accidents are more of a wave on the road ahead. After the big coalition war starts, automatic driving is not a company's business. Now it is an allied period.
2) This is more than an input. It is a charisma, and it is also a long-term combat capability.
Lei Feng Network (search "Lei Feng Net" public concern) Note: This article debut author public number: Automotive Electronics Design, reproduced please contact the authorization and retain the complete information, not to delete the content.