Title: Unveiling the Artificial Mental Model: Enhancing the Passenger Agent’s Decision-Making Process
In the world of artificial intelligence and autonomous systems, the development of sophisticated decision-making models is crucial for ensuring efficient and reliable performance. One such area of focus is the passenger agent, which requires an artificial mental model to navigate the complex transportation landscape. Drawing inspiration from Marvin Minsky’s general purpose mental model, this article explores an innovative approach to designing an artificial mental model for the passenger agent. By incorporating three levels and two layers, this model aims to enhance the decision-making capabilities of the passenger agent, resulting in optimal travel experiences for passengers.
The Artificial Mental Model:
Fig. 1 provides an overview of the proposed artificial mental model for the passenger agent. Comprised of three levels (Reactive, Deliberative, and Reflective) and two layers (Executive and Memory), this model strives to replicate human-like decision-making processes. The Reactive Level serves as the initial response to internal and external stimuli, while the Deliberative Level handles more complex situations. The Reflective Level enables deep reflection on decision-making processes. The Executive Layer classifies stimuli and identifies opportunities and threats, while the Memory Layer manages information storage and retrieval.
To effectively implement the artificial mental model, a mathematical formulation is necessary. In this work, a simulated artificial city is considered, consisting of four types of streets: normal, unsafe, congested, and tourist. The travel time, distance, cost, and utility are calculated based on the characteristics of these streets and other parameters. Various equations, such as those represented by Eq. (1), (2), (3), and (4), enable the quantification of these metrics. The incorporation of a somatic index, as described in Eq. (6), enhances the model’s ability to evaluate and register passenger travel experiences.
Somatic Reaction Behavior:
The article further explores the behavior of somatic reactions by passengers using simulated data. Fig. 2 illustrates the relationship between the intensity of values and the corresponding somatic reactions. By observing the stabilization of reactions as values become more extreme, the article highlights the control and flexibility achieved through the incorporation of somatic reactions. Confidence bands are depicted, representing the randomness effect, which adds a degree of variability to the somatic reaction value.
Somatic Index Valuation:
The valuation of the somatic index is of paramount importance in this research. A neutral initial value is assigned to represent the passenger’s initial experience with a specific transport operator. The most recent somatic reaction experienced by the passenger positively or negatively influences the update of the somatic index. Eq. (7) defines this relationship, emphasizing the autoregressive effect in which prior somatic indexes impact subsequent observations. The artificial somatic reward or punishment associated with the most recent experience influences the passenger’s decision-making process.
General Trip Request Process:
Algorithm 1 elucidates the general trip request process of a passenger agent within the defined artificial mental model. With steps categorized into executive processes, memory management, specific solving mechanisms, and decision-making components, this algorithm presents a comprehensive approach to handling trip requests. Notably, the algorithm incorporates real-time route replanning, travel metric calculation, and the recording of somatic reactions and indexes, ultimately enhancing the passenger agent’s decision-making abilities.
The development of an artificial mental model for the passenger agent showcases the potential for improving decision-making processes within autonomous transportation systems. By leveraging Marvin Minsky’s mental model and incorporating mathematical formulations and somatic reactions, the passenger agent can achieve more efficient and personalized travel experiences. Future research may delve into refining the model and exploring its application in real-world scenarios, ultimately paving the way for smarter, more intuitive passenger agents in the transportation landscape.