Synergising Human-like Responses and Machine Intelligence for Planning in Disaster Response

TL;DR

An attention-based cognitive architecture inspired by Dual Process Theory is proposed that integrates, in an online fashion, rapid yet heuristic (human-like) responses (System 1) with the slow but optimized planning capabilities of machine intelligence (System 2).

Abstract

In the rapidly changing environments of disaster response, planning and decision-making for autonomous agents involve complex and interdependent choices. Although recent advancements have improved traditional artificial intelligence (AI) approaches, they often struggle in such settings, particularly when applied to agents operating outside their well-defined training parameters. To address these challenges, we propose an attention-based cognitive architecture inspired by Dual Process Theory (DPT). This framework integrates, in an online fashion, rapid yet heuristic (human-like) responses (System 1) with the slow but optimized planning capabilities of machine intelligence (System 2). We illustrate how a supervisory controller can dynamically determine in real-time the engagement of either system to optimize mission objectives by assessing their performance across a number of distinct attributes. Evaluated for trajectory planning in dynamic environments, our framework demonstrates that this synergistic integration effectively manages complex tasks by optimizing multiple mission objectives.

Figures (7)

• Figure 1: Rapid responses (System 1), and rational reasoning (System 2) in human cognition.
• Figure 2: The figure illustrates the proposed tripartite cognitive architecture.
• Figure 3: The figure illustrates the dynamic behavior of the attention parameters $\xi_1(t), ~\xi_2(t),$ and $\xi_3(t)$ based on the following configuration $w_{1a}=40$, $w_{1b}=0.02$, $w_{1c}=9$, $w_{1d}=1$, $w_{2a}=10$, $w_{2b}=0.05$, $w_{2c}=7$, $w_{2d}=1$, $w_{3a}=10$, $w_{3b}=0.15$, $w_{3c}=7$, and $w_{3d}=1$.
• Figure 4: The figure illustrates the proposed approach in a simulated disaster scenario, demonstrating the operation of the proposed tripartite cognitive architecture. (a)-(f) Trajectory planning using a combination of System 1 and System 2 based on attribute performance, (g) The evolution of attention parameters during the mission, (h) The activation of different attributes based on the attention vector $A_t$.
• Figure 5: The figure illustrates the probability distribution on the attention probabilities $(p_1,p_2,p_3)$ i.e., Eq. \ref{['eq:dir']} assigned to the 3 attributes at different points in time during the mission.