Editor's introduction: Some partners have some understanding of the knowledge of DMS, but they still do not know where to start regarding the specific landing of the product. Therefore, the author starts from his own practical experience and shares specific product cases about DMS, mainly from the aspects of writing DMS product PRD and planning, hoping to help you.
A friend said in a private message that although he has a certain understanding of DMS, he still does not know how to start to be responsible for the specific landing of the product.
Today, I share the specific product cases I have done about DMS in the main engine factory (automobile factory).
Based on my own experience, this article shares two parts:
First of all, we know that DMS products are system products that integrate hardware and software.
Main parts: camera module (including lens and bracket), dms controller, infrared fill light.
The camera and fill light are connected to the dms controller via a wire-end interface such as the fpd-link iii interface.
Cockpit Domain Controller: Provides audible prompts and interactive interfaces
Gauges: Provide fatigue driving warning icon reminders
The above hardware system and part relationship, the system engineer to complete the specific program design and output sor document. The amount of involvement of the product manager in this process is determined by the actual situation of the company and the individual's ability.
However, the prd of the product software must be done by the product manager himself.
At present, cooperation with third parties is the DMS development model of the vast majority of car companies, and I take this scenario as an example to provide a PRD reference for DMS.
As explained above, the DMS function of this project is completed in cooperation with the supplier. In this collaboration, the vendor provides an algorithm sdk to complete the data analysis of the image and the logical inference of the behavior. However, the judgment criteria for driver behavior inference are configured by us to make specific parameters. The prd will focus on writing the criteria and degree of driver fatigue (distraction, risky behavior).
Taking the fatigue test chapter as an example, at least the following points need to be explained:
(1) Functional description of fatigue detection
During the driving process, the camera samples the driver's closed eyes and yawning behavior; dms combines driving time, driving speed and other factors to determine whether the driver is tired and fatigue level. The system issues a corresponding warning to the driver according to the fatigue level.
Function on/off user portal (interface design)
Set up the interface through the car control
Turned off/on by voice
Fatigue testing is initiated under conditions
车速 30km/h-250km/h
The Law Detection function is turned on
The level and judgment logic of fatigue detection
The fatigue level is set to primary fatigue and secondary fatigue, and the trigger is alarmed, and a high sensitivity algorithm is used.
After the fatigue detection function is started, if the host system receives a first-level fatigue signal, it immediately broadcasts a tts reminder, at which time the fatigue detection continues to execute, and when it meets the second-level fatigue, the dms system has a signal of second-level fatigue. After the system receives a moderate fatigue signal for 3s (about the tts of the mild fatigue reminder have just finished playing), a secondary fatigue reminder is performed.
(2) Fatigue reminder (human-computer interaction)
Primary fatigue: The following are broadcast randomly by tts:
Secondary fatigue: The fatigue tone is emitted first; at the same time, the fatigue driving icon or text is displayed on the instrument, showing 5s hidden.
After the fatigue tone ends, tts randomly broadcasts the following:
Minimum reminder interval: 2 minutes
Like interface and human-computer interaction design, assigned to ued colleagues to complete.
Product planning needs to be based on business planning, business planning needs to have a deep understanding of the business, including a deep understanding of products, technology, management, financial planning, etc. based on the industry. If ims' business is well understood, knows the direction, and knows the rhythm, then the product planning is to pull out the radish and bring out the mud.
IMS is a smart cockpit monitoring system and one of the two major AI areas of smart cockpit, but many companies are not yet aware of the importance of their product planning. Like ai voice systems, if you want to make product differences, you must invest more self-research resources.
Intelligent visual system and intelligent voice system, many enterprises as a functional product to develop, its product team, project team only focus on the development and landing of functional attributes, this idea is very narrow.
In my eyes, whether it is intelligent voice or intelligent vision system, they are platform-based and service-oriented products. The reason for this is that in addition to providing users with the functions of voice interaction and visual monitoring that solve user needs, they also need to provide service frameworks for other functions to meet the intelligent experience of various car use scenarios.
For example, the following scenario:
"The morning of October 1st. The man drove, carrying his wife and 5-year-old son, from Nanshan, Shenzhen, to The Guangzhou Chimelong Zoo. The boy was irritable and restless on the way, got up and stood, unwilling to continue sitting upright. ”
Suppose, we design a function to help parents comfort their children and make them sit down obediently.
First, the cockpit needs to recognize the scene (relying on the oms camera and visual algorithms, navigation information); second, the cockpit needs to actively initiate interaction with the child (relying on voice interaction); and then, ensure that the content and form of the interaction are in line with the child's interests (relying on content services).
It can be seen that in the process of realizing this function in this scenario, intelligent vision and intelligent voice provide service capabilities.
The above is just a scenario.
In actual life, there are layers of continuous scenes and user needs to be tapped, intelligent voice and intelligent vision for different layers of scenes, should build a set of standards and interfaces, with platform rules to serve different scene functions.
Therefore, intelligent vision and intelligent voice must be planned with platform-based and service-oriented products, and through the in-depth mining of car scenes, provide functions and services that meet various car use scenarios, in order to build the differentiation and competitiveness of product experience.
From the enterprise level, if you want to make a difference in the direction of product intelligence, you should increase investment in the self-research of core algorithms in the two AI fields of intelligent vision and intelligent voice, and master the data, get rid of excessive dependence on suppliers, and get rid of business bottlenecks from suppliers.
The specific analytical materials are not shown, and the following directly says the conclusion:
Status: Active DMS systems have become the industry standard
Trend 1: Dms and oms are integrated into cockpit monitoring systems, providing a technical basis for the unsound functions of the cockpit
Trend 2: The visual integration of the inside and outside of the vehicle with other sensing data, and the integration of safety inside and outside the vehicle
Extract key - such as algorithm business:
Extract the key-algorithm software architecture:
The visual processing algorithm module is divided into computing power adaptation layer, data layer, visual perception layer, logic layer, and interface layer, so that layering can improve the reuse of software, the collaboration between modules, and the rational use of hardware platform computing power.
Output function roadmap (schematic form only):
Based on the above, the planning material for the output product is produced.
Because of the limited requirements of the document, it cannot be fully displayed, and I hope that the above content can provide a reference perspective for readers and friends.
This article was originally published by @Cyber Seven on Everyone is a Product Manager, and reproduction without permission is prohibited.
The title image is from unsplash and is based on the cc0 protocol