Problem Space Research Articles

AI-driven automated discovery tools reveal diverse behavioral competencies of biological networks.

Many applications in biomedicine and synthetic bioengineering rely on understanding, mapping, predicting, and controlling the complex behavior of chemical and genetic networks. The emerging field of diverse intelligence investigates the problem-solving capacities of unconventional agents. However, few quantitative tools exist for exploring the competencies of non-conventional systems. Here, we view gene regulatory networks (GRNs) as agents navigating a problem space and develop automated tools to map the robust goal states GRNs can reach despite perturbations. Our contributions include: (1) Adapting curiosity-driven exploration algorithms from AI to discover the range of reachable goal states of GRNs, and (2) Proposing empirical tests inspired by behaviorist approaches to assess their navigation competencies. Our data shows that models inferred from biological data can reach a wide spectrum of steady states, exhibiting various competencies in physiological network dynamics without requiring structural changes in network properties or connectivity. We also explore the applicability of these 'behavioral catalogs' for comparing evolved competencies across biological networks, for designing drug interventions in biomedical contexts and synthetic gene networks for bioengineering. These tools and the emphasis on behavior-shaping open new paths for efficiently exploring the complex behavior of biological networks. For the interactive version of this paper, please visit https://developmentalsystems.org/curious-exploration-of-grn-competencies.

AI-driven automated discovery tools reveal diverse behavioral competencies of biological networks

Many applications in biomedicine and synthetic bioengineering rely on understanding, mapping, predicting, and controlling the complex behavior of chemical and genetic networks. The emerging field of diverse intelligence investigates the problem-solving capacities of unconventional agents. However, few quantitative tools exist for exploring the competencies of non-conventional systems. Here, we view gene regulatory networks (GRNs) as agents navigating a problem space and develop automated tools to map the robust goal states GRNs can reach despite perturbations. Our contributions include: (1) Adapting curiosity-driven exploration algorithms from AI to discover the range of reachable goal states of GRNs, and (2) Proposing empirical tests inspired by behaviorist approaches to assess their navigation competencies. Our data shows that models inferred from biological data can reach a wide spectrum of steady states, exhibiting various competencies in physiological network dynamics without requiring structural changes in network properties or connectivity. We also explore the applicability of these ‘behavioral catalogs’ for comparing evolved competencies across biological networks, for designing drug interventions in biomedical contexts and synthetic gene networks for bioengineering. These tools and the emphasis on behavior-shaping open new paths for efficiently exploring the complex behavior of biological networks. For the interactive version of this paper, please visit https://developmentalsystems.org/curious-exploration-of-grn-competencies.

DaedalusData: Exploration, Knowledge Externalization and Labeling of Particles in Medical Manufacturing - A Design Study.

In medical diagnostics of both early disease detection and routine patient care, particle-based contamination of in-vitro diagnostics consumables poses a significant threat to patients. Objective data-driven decision-making on the severity of contamination is key for reducing patient risk, while saving time and cost in quality assessment. Our collaborators introduced us to their quality control process, including particle data acquisition through image recognition, feature extraction, and attributes reflecting the production context of particles. Shortcomings in the current process are limitations in exploring thousands of images, data-driven decision making, and ineffective knowledge externalization. Following the design study methodology, our contributions are a characterization of the problem space and requirements, the development and validation of DaedalusData, a comprehensive discussion of our study's learnings, and a generalizable framework for knowledge externalization. DaedalusData is a visual analytics system that enables domain experts to explore particle contamination patterns, label particles in label alphabets, and externalize knowledge through semi-supervised label-informed data projections. The results of our case study and user study show high usability of DaedalusData and its efficient support of experts in generating comprehensive overviews of thousands of particles, labeling of large quantities of particles, and externalizing knowledge to augment the dataset further. Reflecting on our approach, we discuss insights on dataset augmentation via human knowledge externalization, and on the scalability and trade-offs that come with the adoption of this approach in practice.

Systematic refinement of surrogate modelling procedure for useful application to building energy problems

Systematic procedures for applying surrogate model development to building energy problems have limited adoption and struggle to incorporate advancements in broader machine learning research. This work demonstrates an iterative approach that encompasses: Establishing clear, consistent baseline performance thresholds. Assembling a comprehensive set of domain-relevant evaluation metrics, including rigorous bounds on error and emphasis on regions of interest within the broader problem space. Characterizing and adapting the problem space using sub-sampling and pre-processing techniques. Accounting for variability, randomness and complexity in the building energy problem definition, data sampling and surrogate model training. Refining the metamodel decision space defining the neural network architecture and training algorithms, and explored by hyperparameter optimization methods. In addition to demonstrating improved performance predicting aggregate annual heating and cooling demands for an illustrative office case, accuracy, bias and bounds on error were all brought within domain-relevant thresholds for net zero regions of interest. Highlights Iterative refinement of surrogate modelling procedure for building energy problems Consideration of complexity and variability in energy data, problem and surrogate Addition of key dimensions to performance evaluation useful for domain application Greatly improved predictive value for net zero building energy regions of interest Systematic shift in hyperparameter tuning towards deep learning configurations

Stopper vs. Singular Controller Games With Degenerate Diffusions

We study zero-sum stochastic games between a singular controller and a stopper when the (state-dependent) diffusion matrix of the underlying controlled diffusion process is degenerate. In particular, we show the existence of a value for the game and determine an optimal strategy for the stopper. The degeneracy of the dynamics prevents the use of analytical methods based on solution in Sobolev spaces of suitable variational problems. Therefore we adopt a probabilistic approach based on a perturbation of the underlying diffusion modulated by a parameter γ>00$$\\end{document}]]>. For each γ>00$$\\end{document}]]> the approximating game is non-degenerate and admits a value uγ and an optimal strategy τ∗γ for the stopper. Letting γ→0 we prove convergence of uγ to a function v, which identifies the value of the original game. We also construct explicitly optimal stopping times θ∗γ for uγ, related but not equal to τ∗γ, which converge almost surely to an optimal stopping time θ∗ for the game with degenerate dynamics.

Point-by-point transfer learning for Bayesian optimization: An accelerated search strategy

Bayesian optimization (BO) is a prominent “black-box” optimization approach. It makes sequential decisions using a Bayesian model, usually a Gaussian process, to effectively explore the search space of laborious optimization problems. However, BO faces notable challenges, particularly in constructing a reliable model for the optimization task when there are insufficient data available. To address the “cold start” problem and enhance the efficiency of BO, transfer learning appears as a powerful strategy which has gained notable attention recently. This approach aims to expedite the optimization process for a target task by utilizing knowledge accumulated from previous, related source tasks. We provide a novel point-by-point transfer learning with mixture of Gaussians for BO (PPTL-MGBO) technique to improve the speed and efficacy of the optimization process. Through evaluations on both synthetic and real-world datasets, PPTL-MGBO has demonstrated marked advancements in optimizing search efficiency, particularly when dealing with sparse or incomplete target data.

Supporting multi-criteria decision-making processes with unknown criteria weights

Decision support systems are crucial in today’s tech-driven world, assisting decision-makers with complex choices. Determining criteria weights is a paramount aspect, significantly influencing outcomes. Traditionally, criteria weights are derived from objective measures, subjective expert knowledge, or a combination of both, each with its own strengths and limitations. This paper presents a novel approach for addressing unknown criteria relevance by systematically generating weight vectors, thus exploring a broader decision problem space. The proposed methodology is adaptable to various multi-criteria methods, enhancing its applicability across different scenarios. Its effectiveness is empirically validated through two practical examples: Glomerular Filtration Rate (GFR) evaluation and bridge construction method selection, demonstrating its broad applicability. Comparative analysis with existing objective weighting techniques reveals the limitations of current approaches and highlights the improved decision-making capabilities enabled by the proposed method. This research addresses a critical gap in the reliability and robustness of existing methods, particularly in situations with unknown criteria weights. Key contributions include a new decision-making methodology and an innovative ranking formulation using fuzzy sets, with empirical verification strengthening the utility of the approach. This paper offers a promising solution for advancing multi-criteria decision analysis, especially in complex scenarios with uncertain criteria relevance.

When cities broke into the global stage: 20 years since the publication of ‘Cities and Climate Change’

This short symposium revisits the book Cities and Climate Change by Harriet Bulkeley and Michelle Betsill and its role in establishing cities as central actors in climate change governance. The contributors to this symposium reflect on the research themes and the book’s legacy from their disciplinary perspectives. The first response explores the tensions between the elevation of cities as sites for climate action and the persistent lack of state action, arguing for meaningful public participation and a labour of making visible how city-based practices respond to climate change. The second response examines how the book has enabled generations of scholars to appreciate the complex constitution of city actors and how they interact/enact global climate governance yet remain frustratingly marginal in the discipline of International Relations. The third response explores how the book positioned the urban as a problem space, a manageable scale of intervention, and a political space with the capacity to enact innovation, triggering new questions about the potential and limits of the expanding responsibilities of cities. The fourth response argues that as cities become active in adaptation, they need to engage in an anticolonial politics of love and hospitality that sheds ‘security’ and makes the self vulnerable to ‘others’ for an effective response to climate change. The symposium concludes with a recap by Betsill and Bulkeley which underscores the variety of interdisciplinary debates that the book has generated, and its influence on current thinking on debates on climate change governance.

Human and Machine: A View through Kant's Use of Regulative Reason

This study examines a qualitative difference between human intelligence and artificial intelligence (AI) through the lens of Immanuel Kant's philosophy. This difference is based on the human mind's idea of unity and its intuitive ability to limit this unity. Kant argues that through a regulative use of reason, it is able to obtain conceptual wholes such as God, soul, and physics. We argue that this idea of totality, which Kant obtains through the use of regulative reason, has an important function in distinguishing the human mind from artificial intelligence. Based on Kant's idea of unity, our study determines that the human mind constructs a unity like a formal system. Artificial intelligence, which is made possible by this construction, is inadequate in solving many problems that have shown surprising developments in recent years. We attribute this inadequacy to the inability of computers to model the aforementioned idea of wholeness. We justify this claim through the problem of “narrowing the brute search space” in computational complexity, which is a problem area in computer science. This problem is based on the lack of a formal procedure for narrowing down a problem space with very large boundaries. When computers do not have an efficient procedure or an analytical solution to the problems they want to solve, they are forced to try all available solutions. In contrast, the human mind has cognitive abilities that allow it to intuitively narrow down these large problem spaces. Kant's use of regulative reason provides a framework for understanding this human faculty. According to Kant, the regulative function of reason provides the concepts of pure reason that orient scientific inquiry. For example, through a concept of pure reason, such as the cosmos, the human mind is able to limit the physical domain in such a way as to do physical science. Through Kant's use of regulative reason, we think that the human mind, by bringing together a set of formal signs, makes a limitation such as a formal system. Since we can conceptualize a unity such as a formal system, we can talk about algorithms that operate according to this system. The regulative use of reason, which enables the establishment of such wholes, creates a qualitative difference between AI when combined with the human's intuitive thinking ability. However, it is quite difficult to ground the connection between the idea of unity in Kant's philosophy and intuitive thinking solely on the basis of Kantian philosophy. Thus, we turn to the views of Henri Bergson and Nazif Muhtaroğlu to establish this connection. Bergson, while explaining the concept of motion, argues that the mind reaches such an idea of unity through an instinctive synthesis. In this respect, Bergson argues that movement is a mental synthesis insofar as it is a transition from one point to another. Similarly, Muhtaroğlu, after emphasizing that the intuitive cognition that accompanies reason is a direct, unmediated and rapid cognition, identifies the type of intuition that leads to the idea of unity as immediate intuition. Stating that this type of intuition is a cognitive intuition, Muhtaroğlu cites Archimedes' discovery of the laws of fluids as an example of this way of thinking. In our study, we use the example of Archimedes to show how intuition accompanies the narrowing of the field of brute force search. Thus, when the regulative use of reason and intuitive thinking come together, a difference emerges in the cognitive abilities of the human mind and artificial intelligence. Thanks to the regulatory use of reason, the human mind is able to have an awareness of the unity of the object field it is confronted with. The fact that this awareness is accompanied by intuitive thinking allows this field of unity to be narrowed. Since artificial intelligence cannot model both the use of regulative reasoning and intuitive thinking, it is subjected to the brute search method. We argue that such a deficiency underlies the lack of analytical solutions to problems of computational complexity. This deficiency reveals the difference between the human mind and artificial intelligence in problem solving and narrowing down large search spaces.

Integrated Workflow for High-Throughput Formulation and Characterization of Battery Electrolytes

Developing novel electrolyte formulations is a major challenge to the adoption of new electrode chemistries in electrochemical energy storage devices. Electrolyte engineering is a massive field, with hundreds of possible solvents, salts, and additives to choose from. The many building blocks lead to an immense combinatorial problem and enormous parameter space for researchers to search within. Computational tools are a promising solution to the issue, with models that can predict electrolyte properties using from machine learning,1 thermodynamic calculations,2 and molecular dynamics.3 Though incredibly powerful, these calculations can be very computationally expensive, highly specific to a certain subset of electrolytes and often require experimental training data or validation. The availability of electrolyte datasets is currently limited to computational or small experimental datasets, and the absence of standardization in the field makes it challenging to extract a cohesive dataset from the literature. Therefore, there is a need for experimental tools to generate new electrolyte datasets for powerful data analysis techniques to enable novel battery chemistries.In this presentation, we will discuss an automated system for electrolyte formulation and high throughput characterization of electrochemical stability, Coulombic efficiency, and ionic conductivity. Existing automated systems lack solid dispensing and are thus limited in the formulation space that is accessible by automation.4,5 Our design utilizes a robotic arm to handle and dispense solid and liquid precursors to formulate electrolyte samples and heating and stirring capabilities for electrolyte mixing. Post formulation, the robotic platform utilizes custom-made characterization cells to facilitate high-throughput analysis. Specifically, we have designed a microplate-style Coulombic efficiency testing cell and a series of flow-through sensors to collect ionic conductivity and electrochemical stability information that only utilizes a small electrolyte volume. Initial studies with the platform target aqueous electrolytes to enable ease of troubleshooting outside of a glovebox. Here we screen a variety of salts including LiOAc, LiNO3, LiSO4, LiTFSI and LiFSI. The chosen salts contain anions spanning a wide range of the Hofmeister series, which classifies them between chaotropic (structure breaking) and kosmotropic (structure making). Chaotropic anions, such as TFSI- and FSI-, affect the water bonding structure, disrupting it, leading to an increase in the electrochemical stability window of the electrolyte.6 We find that we can explore and expand this phenomenon in combinations of the lithium salts in our high-throughput system.We aim to leverage this high-throughput electrolyte characterization platform to produce high quality, consistent, open-source databases of electrolyte properties that we envision will assist and accelerate the entire field’s research. Our long-term goal is to use the comprehensive data generated to make fundamental advances in developing new electrolyte models that will allow researchers to quickly predict optimal formulations and device performance.1. S. C. Kim et al., Proceedings of the National Academy of Sciences, 120, e2214357120 (2023).2. A. Dave, K. L. Gering, J. M. Mitchell, J. Whitacre, and V. Viswanathan, J. Electrochem. Soc., 167, 013514 (2019).3. B. Ravikumar, M. Mynam, and B. Rai, J. Phys. Chem. C, 122, 8173–8181 (2018).4. A. Dave et al., arXiv:2111.14786 [cs] (2021) http://arxiv.org/abs/2111.14786.5. S. Matsuda, K. Nishioka, and S. Nakanishi, Sci Rep, 9, 6211 (2019).6. D. Reber, R. Grissa, M. Becker, R.-S. Kühnel, and C. Battaglia, Advanced Energy Materials, 11, 2002913 (2021).

State discretization for continuous-state MDPs in infectious disease control

Repeated decision-making problems may arise in the health policy context, such as infectious disease control for COVID-19 and other epidemics. These problems may sometimes be effectively solved using Markov decision processes (MDPs). However, the continuous or large state space of such problems for capturing infectious disease prevalence renders it difficult to implement tractable MDPs to identify the optimal disease control policy over time. We therefore develop an algorithm for discretizing continuous states for approximate MDP solutions in this context. We benchmark performance against a uniform discretization using both a synthetic example and an example of COVID-19 in Los Angeles County.

Recovering African contestation and innovation in global politics: Francis Deng and sovereignty-as-responsibility

Abstract Against both liberal narratives and postcolonial critiques, this article argues that sovereignty-as-responsibility – the theory of sovereignty embraced in the responsibility to protect (R2P) – is part of a problem space that emerged with decolonization, rather than the end of the Cold War. The internally displaced person (IDP), the vehicle which Francis Deng used to critique Westphalian sovereignty, had to be theorized against the rise of the postcolonial state. In recovering the questions motivating Deng we find a stark politics driving his work on IDPs and sovereignty. Against the claim that the heart of R2P is armed coercive intervention for humanitarian purposes, Deng used sovereignty-as-responsibility to promote a profoundly political critique of the colonial legacy and the postcolonial state, which was taken up by states of the Global South in debates on the ratification of R2P. Recovering Deng's work on IDPs and sovereignty-as-responsibility highlights R2P as itself a site of contestation, and offers a case for how ideas emerge ‘from below’ in global politics.

Muddy Waters: Design Thinking for Understanding the Multi-Organizational Problem Space of the Water Sector

Context and motivation: Climate change is manifested by climate variability, rising temperatures (and thus evaporation), and extreme events such as droughts and floods, which have a profound effect on the availability of natural resources, for example, high-quality water. While several technologies for addressing these challenges are available, their adoption is not widespread. In this study, a design thinking (DT) approach was applied to understand the problem space of floods and their handling by the Israeli water sector. Specifically, we aim at addressing the following question: What are the gaps in and barriers to adopting solutions that address sewerage flooding during extreme heavy rainfall events? The DT approach exposed major problems in the conduct of the water sector, including a lack of communication among organizations, the ill-defined distribution of responsibility, unclear and conflicting guidance, and insufficient funds and technological solutions, all hindering the possibility of adopting an integrative solution. This study demonstrates the role that DT plays in understanding a complex, multi-organizational problem space, in our case, the climate change readiness of the water sector, before delving into technological development. Any solution development should involve participants from the various organizations involved in the challenge. It is vital to address not only each organization’s requirements but also its technology adoption barriers and to initiate a comprehensive discussion, ultimately resulting in a shared understanding of all the facets of the challenge that can impact solution development and deployment.

Reconsidering Divergent and Convergent Thinking in Creativity – a Neurophysiological Index for the Convergence-Divergence Continuum

ABSTRACT Creative problem-solving has been described as an iterative process of divergent and convergent modes of thinking. With this opinion paper, we illustrate how some tasks, commonly used for the neuroscientific investigation of creativity might provide a condensed representation of divergent and convergent thinking, respectively. We have identified the following reasons for this: The specific presentation of stimuli and response alternatives within these tasks (“immediate problem space accessibility”), the restricted time on task for participants in well-established creativity tasks, as well as the challenges to identify specific temporal dynamics of divergent-convergent cycles using standard EEG procedures. Based on two recent studies, we propose that upper alpha band (10–12 hz) modulations in particular may serve a) as a neuro-physiological index to disentangle divergent-convergent cycles within these creativity tasks and b) as a data-driven marker for the task demands related to immediate problem space accessibility. Finally, we argue for a view on divergent and convergent thinking that is less of a dichotomy and more of a continuum.

Novel PCA-Based Lower-Dimensional Remapping of the Solution Space for a Genetic Algorithm Optimization: Estimating the Director Distribution in LC-Based SLM Devices

This work introduces a novel computational approach based on Principal Component Analysis (PCA) for dimensionality reduction of the solution space in optimisation problems with known linear interdependencies among solution variables. By creating synthetic datasets with deliberately engineered properties and applying PCA, the solution space’s remapping significantly reduces its dimensionality, leading to faster computation and more robust convergence in optimisation processes. We demonstrate this method by integrating it with a Genetic Algorithm (GA) for solving the optimal director distribution in liquid crystal (LC) devices, specifically addressing 2D and complex 3D spatial light modulator (SLM) structures such as twisted nematic liquid crystals (TN-LC) and parallel-aligned liquid crystal on silicon (PA-LCoS), respectively. The phase profiles obtained from the director vector distributions for horizontal and vertical high-frequency binary phase gratings closely match the theoretical values derived from minimising the traditional elastic Frank–Oseen functional via Euler–Lagrange equations. Beyond this specific application, our method offers a general framework for reducing computational complexity in optimisation problems by directly reducing the dimensionality of the solution space. This approach is applicable across various optimisation scenarios with well-known linear interdependencies among solution variables, enabling significant reductions in computational costs and improvements in robustness and convergence.

Prompt Update Algorithm Based on the Boolean Vector Inner Product and Ant Colony Algorithm for Fast Target Type Recognition

In recent years, data mining technology has become increasingly popular, evolving into an independent discipline as research deepens. This study constructs and optimizes an association rule algorithm based on the Boolean vector (BV) inner product and ant colony optimization to enhance data mining efficiency. Frequent itemsets are extracted from the database by establishing BV and performing vector inner product operations. These frequent itemsets form the problem space for the ant colony algorithm, which generates the maximum frequent itemset. Initially, data from the total scores of players during the 2022–2024 regular season was analyzed to obtain the optimal lineup. The results obtained from the Apriori algorithm (AA) were used as a standard for comparison with the Confidence-Debiased Adversarial Fuzzy Apriori Method (CDAFAM), the AA based on deep learning (DL), and the proposed algorithm regarding their results and required time. A dataset of disease symptoms was then used to determine diseases based on symptoms, comparing accuracy and time against the original database as a standard. Finally, simulations were conducted using five batches of radar data from the observation platform to compare the time and accuracy of the four algorithms. The results indicate that both the proposed algorithm and the AA based on DL achieve approximately 10% higher accuracy compared with the traditional AA. Additionally, the proposed algorithm requires only about 25% of the time needed by the traditional AA and the AA based on DL for target recognition. Although the CDAFAM has a similar processing time to the proposed algorithm, its accuracy is lower. These findings demonstrate that the proposed algorithm significantly improves the accuracy and speed of target recognition.

Common errors in Generative AI systems used for knowledge extraction in the climate action domain

Large Language Models (LLMs) and, more specifically, the Generative Pre-Trained Transformers (GPT) can help stakeholders in climate action to explore digital knowledge bases, extract and utilize climate action knowledge in sustainable manner. However, LLMs are "probabilistic models of knowledge bases" that excel at generating convincing texts but cannot be entirely relied upon due to the probabilistic nature of the information produced. This brief report illustrates the problem space by shedding some light on the issues of incomplete answers, hallucinations, and misinformation threats.

Who is the entrepreneurial educator? A transferability perspective to tease out antecedents needed to form a signature pedagogy

Purpose Entrepreneurial education (EE) has grown rapidly and become important for how we prepare future generations for work. However, a less addressed piece of the puzzle is: who is the entrepreneurial educator? In this paper, our aim draws upon Palmer’s (1998) idea of wholeness, where we seek to connect the two problem spaces of the role diversity of entrepreneurial educators and how the educator is closely tied to the evolutionary development of the domain in constructing signature pedagogies that could cater for the issue of a low capitalization rate.Design/methodology/approach We problematize on a conceptual level the entry-level personas that have been addressed in prior literature by using the processes of transferability and evolutionary theory to tease out how variability among educators could create antecedents to form a signature pedagogy.Findings We recognize four archetypes: the experienced entrepreneur, the entrepreneurship scholar, the academic scholar and the pedagogical expert. Our problematization on the variance of the four archetypes provides a fertile ground to engage in addressing “the elephants in the room” in EE, the capitalization rate of graduates and the importance of developing a signature pedagogy, acknowledged in other domains, to enhance legitimacy.Originality/value The paper provides a nuanced outlook on who is teaching in the entrepreneurial classroom and problematizes from an evolutionary perspective the potential future paths to continue staying relevant for society as well as justifying our presence as a legit domain in academia.

Doubly special relativity as a nonlocal quantum field theory

In this work, we explore the quantum theories of the free massive scalar, the massive fermionic, and the electromagnetic fields in a doubly special relativity scenario. This construction is based on a geometrical interpretation of the kinematics of this kind of theory. In order to describe the modified actions, we find that a higher (indeed infinite) derivative field theory is needed, from which the deformed kinematics can be read. From our construction, we are able to restrict the possible models of doubly special relativity to those that preserve linear Lorentz invariance. We quantize the theories and also obtain a deformed version of the Maxwell equations. We analyze the electromagnetic vector potential for both an electric pointlike source and magnetic dipole. We observe that the electric and magnetic fields do not diverge at the origin for some models described with an anti–de Sitter momentum space, but they do for a de Sitter space in both problems. Published by the American Physical Society 2024

ICOPP: A Highway Autonomous Driving Planner Based on Interactive Collaborative Optimization and POMDP

Abstract In this paper, we propose a planner for autonomous vehicles in complex highway scenarios, which can efficiently plan the trajectory of autonomous vehicles. The method aims to improve the intelligence of autonomous driving under the uncertainty of the perception and prediction system. The method employs a POMDP framework to address the challenges brought about by uncertainty. Firstly, this method divides the entire planning cycle into several discrete decision-making processes. In each process, the vehicle’s lateral and longitudinal behaviors are semantically processed to obtain a limited number of decision sequences, thereby addressing the curse of dimensionality in state space and observation space in POMDP problems. In each decision-making process, different simulation scenarios are constructed based on the decision of the self-vehicle and probability distribution of the predicting behaviors of surrounding vehicles. In each scenario, key vehicles are selected for collaborative optimal planning among multiple vehicles, resulting in the optimal drivable trajectories for both the ego vehicle and other vehicles under the given decision. Finally, an evaluation of the policy is conducted in terms of safety, comfort, and efficiency to select the optimal policy. In simulation experiments, the ICOPP demonstrates smoother and more stable autonomous driving performance compared to traditional methods.