Author: Zen and the Art of Computer Programming
1 Introduction
In recent years, with the continuous development of artificial intelligence technology, many industries have begun to transform to be driven by artificial intelligence instead of traditional industrial models. Among them, wearable products and virtual reality technology (VR/AR) have become a trend. Current artificial intelligence systems are mainly implemented based on rules, algorithms, data, etc. How to achieve a better interactive experience and develop high-quality robots and machine vision products in this new industrial environment is a key issue. Research in this field mainly focuses on two directions: on the one hand, trying to establish users' understanding and cognition of the virtual reality or augmented reality world through machine learning methods, thereby improving the user's sense of reality and immersion; on the other hand, , applying physical human-computer interaction skills to virtual reality and augmented reality scenarios to improve the quality and efficiency of life. However, these technologies often rely on massive computing resources and large-scale data sets, and for individual users, even in commercial-grade consumer products, there is no guarantee of sufficient response speed. At the same time, both R&D companies and service platform providers need to carry out professional design in terms of marketing strategy, product structure, interactive experience, etc. Therefore, how to effectively use artificial intelligence technology in this new industrial environment, while reducing costs and improving product quality has become an important issue. This article will combine the existing technical research results to explain the new trends in MIND model and AI product design, and give some thoughts on its future development direction.
2. Explanation of basic concepts and terms
2.1 MIND model
The MIND model (Minimum Interactive Digital Narrative Model) is based on human diversity needs and creativity, weaving human social experiences and activities into a "digital adventure" movie. It was first proposed by the University of Maryland team in 2019, aiming to explore and define how to guide users to participate in adventure stories that have been circulated for a long time in human history in a way that minimizes input and output. The MIND model includes three levels: (1) Diversified character construction: By giving characters unique identities and personalities, we create diverse characters, allowing users to construct situations that become more natural, coordinated, and complete. (2) Narrative sorting: By sorting out human behavior habits and customs and cultures in different periods and locations, we create clues with distinctive personalities and provide unique plot development and narrative. (3) Situation projection: Make full use of the user’s environment and make users feel immersed in the environment.
2.2 AI product design
AI product design refers to the use of computer vision, machine learning, deep learning and other technologies to improve product functions and brand premiums, and to improve product performance, user satisfaction and market share through automated operations and decision-making. A typical AI product design framework can be divided into five stages: demand analysis, product form design, product prototyping, product iteration optimization and continuous product improvement.
2.2.1 Case Analysis—AR Glasses
In 2018, Microsoft launched HoloLens, which introduced a human perspective into the virtual reality world. Compared with traditional static glasses, HoloLens makes the images users see in virtual reality more dynamic, exciting and attractive. However, this product has been difficult to attract users' attention due to its high cost, lack of usability and low user acceptance. Therefore, Microsoft introduced speech recognition functionality in subsequent versions of HoloLens. Users can wake up HoloLens on the device, tell it what function they need, and it will help the user complete the task. In addition, gesture recognition and translation functions have also been launched. Although this product was difficult to implement in some areas, it successfully attracted the attention of consumers and thus changed the landscape of the virtual reality industry.
2.2.2 Case analysis - mobile phone anti-theft function
In order to protect users' personal information and mobile phones, the US federal government launched comprehensive mobile phone locking and password management functions in 2017. Due to the influence of globalization, mobile phone locking functions are quickly coming to the market. However, because users tend to forget their passwords during use, their accounts may be stolen. To this end, the federal government has launched a password management project that intends to use artificial intelligence technology to automatically identify account information on mobile phones and recommend the best passwords. Since the project is expected to cost one billion U.S. dollars and the equipment cost is tens of thousands of yuan, the U.S. government chose a cloud computing platform for deployment. The success of the project attracted attention across the industry and prompted other companies to undertake similar attempts.
3. Explanation of core algorithm principles, specific operating steps and mathematical formulas
The MIND model is designed to be "playable without input". The core idea behind this is to use human potential and intuition to construct stories. Therefore, we need to first understand how this model works.
The MIND model must first answer several "whys", such as: How to allow users to understand our universe? How to make users resonate? How to create an interesting and memorable experience? The MIND model consists of four levels, namely diversified character construction, narrative combing, situational projection and interaction design.
(1) Diversified role construction
The first level is the construction of diverse roles. The characters here can be protagonists, objects, landscapes, or even virtual interactive elements in virtual reality, all to enhance the realism and diversity of the scene. When building your character, there are a few things to consider:
Identity: Creating multiple characters of different races, nationalities, and professions will be more in line with human diversity and personality.
Scenario: Shape the character's image according to the user's environment and target needs, making the user's construction of the situation more natural, coordinated, and complete.
Metaphor: Use metaphors between characters to structure the plot and create surprising emotions.
Portraits: Present the character’s characteristics through portraits, allowing users to have a more objective and accurate understanding of the character.
Senses: Through the character’s visual, auditory, smell and other sensory attributes, users can gain a clear, profound and real experience.
(2) Narrative sorting out
The second level is narrative sorting. The MIND model weaves human social experiences and activities into a "digital adventure" movie, so it is necessary to understand the past historical process of human society, the modernization process, the context of civilization evolution, the politics, economy, social contradictions of today's society, and the complexity of human nature. . When sorting out the narrative, you need to consider the following points:
Time and space: Show the authenticity of the development of human society through the time and place of events and the evolutionary path of historical processes.
Theme: Use different themes to guide users to resonate and create narratives of different interests.
Concept: Cultivate users' thinking and perception abilities through complex human behavioral habits, customs and culture, as well as the political and economic contradictions of modern society.
Symbols: By connecting symbols to human experience, users can become accustomed to them and respond to various scenarios with ease.
(3) Situational projection
The third level is situational projection. Situation projection means that by providing different scenes and locations, users can feel that they are in a virtual space and can explore, enjoy, and create without restriction. In this regard, the following points need to be taken into account:
Space: The layout and atmosphere of different locations will affect users’ perception and experience of the situation.
Form of expression: Different forms of media will affect the user’s sensory experience, including animation, photos, videos, etc.
Immersion: Users may feel tired and frustrated before they are fully engaged. Therefore, placing devices or objects in front of a picture that balances the situation needs to attract the user's attention and feedback.
(4) Interaction design
The fourth level is interaction design. When using virtual reality or augmented reality products, users must control their own behavior and provide evaluation and feedback on the results produced by the system. Therefore, when designing interactions, the following points need to be taken into consideration:
Action rules: Set the right action rules so users can easily control their movements and behaviors.
User preferences: Meet different types of user needs by allowing users to customize device parameters and set personal preferences.
Prompt information: By displaying prompt information on the screen, let users know that the system is running and avoid misoperation.
(5) Explanation of mathematical formulas
The MIND model uses a variety of elements such as geometric shapes, colors, sounds, lines, spatial relationships, animations, and text in its design, and also adds a wealth of mathematical formulas. Let’s take a look at these formulas:
(5.1) Diversified role construction
Euler angle formula:
$θ=atan2(\frac{\text{Oz}-\text{Oy}}{\text{Ox}-\text{Oy}})$
$\text{Ox}$ is the unit vector of the X-axis coordinate, $\text{Oy}$ and $\text{Oz}$ are the unit vectors of the Y- and Z-axis respectively.
(5.2) Narrative sorting out
Conic section formula:
$(x-h)^2+(y-k)^2=\sqrt{(r-\delta)^2+z^2}$
$\text{Xo}(t)=\frac{X_o}{2}+\left[\cos t+\sin t\right]\Delta x,\quad \text{Yo}(t)=\frac{Y_o}{2}+\left[-\sin t+\cos t\right]\Delta y,\quad z=Z_o+c_zt,\quad (0≤t≤2\pi)$
$\text{Xo}$, $\text{Yo}$, $Z_o$ are the center point coordinates of the conic surface of the conic section, $\Delta x$, $\Delta y$ are the conical surface with radius $R$ The horizontal length of , $c_z$ is the height.
Projection mapping formula:
$\text{P}=C_{\theta}\text{L}_p$
$\text{C}_{\theta}$ is the cosine value of the light source direction, and $\text{L}_p$ is the reflection vector.
(5.3) Situational projection
Kalman filter formula:
$\text{x}^{\text{t}}=F_{\text{t}}\cdot\text{x}^{\text{t-1}} + B_{\text{t}}\cdot u^{\text{t}}$
$u^{\text{t}}$ is the control command, $F_{\text{t}}$ is the system state transition matrix, and $B_{\text{t}}$ is the control input matrix.
Atmospheric pressure formula:
$P=-\rho\cdot g\cdot h$
$\rho$ is the air pressure, $g$ is the acceleration due to gravity, and $h$ is the altitude.