This study analyzes Graph Neural Networks (GNNs) for distribution system state estimation (DSSE) by employing an interpretable Graph Neural Additive Network (GNAN) and by utilizing an edge-conditioned message-passing mechanism. The architectures are benchmarked against the standard Graph Attention Network (GAT) architecture. Multiple SimBench grids with topology changes and various measurement penetration rates were used to evaluate performance. Empirically, GNAN trails GAT in accuracy but serves as a useful probe for graph learning when accompanied with the proposed edge attention mechanism. Together, they demonstrate that incorporating information from distant nodes could improve learning depending on the grid topology and available data. This study advances the state-of-the-art understanding of learning on graphs for the state estimation task and contributes toward reliable GNN-based DSSE prediction technologies.
This paper proposes a generalized active classification framework designed to classify the environment of a robot through controlled tactile interaction, with direct application to robotic disassembly of Waste Electrical and Electronic Equipment (WEEE). In this domain, purely vision-based systems frequently fail due to occlusions, dust, and reflective surfaces, making haptic identification an operational necessity. By formulating environment identification as an active sensing problem, we utilize the entropy of learned interaction models and a receding-horizon optimization to select control inputs that maximize information gain from force–torque feedback. Our approach transitions from a passive classifier to an active framework that dynamically probes the environment to resolve belief uncertainty across six distinct tactile classes, demonstrated through the high-precision detection and unfastening of screws using an industrial 6-DOF robotic arm. Experimental results on a consumer microwave demonstrate that the active framework achieves a 51% reduction in detection latency for screws compared to passive baselines and consistently outperforms expert-designed heuristics. Furthermore, the system exhibits robust zero-shot generalization to novel industrial objects, indicating that learned tactile signatures are invariant to the global object context. These results provide a scalable foundation for autonomous disassembly in high-variance environments like electronic waste recycling, while offering a broader methodology for any robotic classification task where control can be leveraged to accelerate information gain.
Aggregate hierarchy, the organization by which microaggregates form progressively larger, structurally distinct macroaggregates, is central to soil stability, governing resistance to erosion and response to disturbance. However, the mechanisms and extent of hierarchical breakdown remain poorly quantified across different soil types and land management. In this study, we addressed this gap by evaluating the stepwise breakdown of soil structure into aggregates, driven by incremental sonication energy, across a range of soils differing in mineral composition and management practices. By applying a quantitative modelling framework, we derived three key parameters: the disruption constant (k₁), reflecting the rate of aggregate breakdown; the dispersion constant (k₂), which describes particle release; and the critical energy threshold (Ecrit), which denotes the transition point between aggregate disruption and full particle dispersion. These parameters were used to evaluate the degree of hierarchy in aggregate breakdown, whereby higher k₁/k₂ ratios signal a pronounced stepwise (hierarchical) disintegration, and ratios near unity indicate direct dispersion into clay-sized particles. Our results indicated that soils enriched in 2:1 phyllosilicate clay minerals, such as Luvisols, exhibited markedly higher k₁/k₂ ratios in larger aggregates, demonstrating a structured, multi-step breakdown process. In contrast, oxide-rich soils like Ferralsols and Andosols typically lacked such hierarchy, dispersing rapidly into smaller fractions, which is consistent with a stronger role of mineral-mineral binding relative to organic-mediated aggregation. In the studied Luvisols, direct seeding was associated with higher stability (higher Ecrit) and a greater degree of hierarchy than conventional tillage, emphasizing the synergistic effects of organic matter input and reduced disturbance on soil structural integrity. These findings highlight the mechanistic roles of distinct pedogenic groups and management practices in controlling aggregate hierarchy and stability. Our study shows that sonication-derived indicators can differentiate not only distinct pedogenic groups but also soil management. These indices can be further developed to provide a quantitative insight into the structural organization that underpins water retention, erosion resistance, and other soil functions critical for conservation.
Material footprints (MF) and the broader framework of Material Flow Accounting (MFA) have gained prominence as indicators of human pressure on the environment, particularly in policy discussions on resource efficiency and circular-economy strategies. While MF and MFA can be useful as a descriptive measure of industrial metabolism, this paper argues that they exhibit several fundamental scientific limitations that distinguish them from other widely used indicators of human demand such as carbon footprints, ecological footprints, or human appropriation of net primary productivity. These limitations arise primarily from the absence of a biophysically grounded aggregation principle and the lack of an intrinsic upper bound. As a result, material footprints are analytically weak and potentially misleading as a sustainability metric, even though they remain valuable as a throughput indicator. To address these limitations, this paper outlines strategies to reduce misuse and deploy material footprint accounting more effectively as a tool for sustainability transitions.
Home appliance companies often offer after-sales services, especially repair services, to their customers. A critical challenge in this context is planning the allocation and routing of service technicians to satisfy customers’ service requests. In this paper, we consider technicians with two different skill levels. Senior technicians with extensive experience consistently handle repair tasks successfully, whereas junior technicians, owing to their limited experience, may fail to repair appliances, necessitating a follow-up visit by a senior technician to complete the service. Stochasticity of the problem stems from random service results and arrival of new customers. We formulate the problem as a sequential decision problem, where at the start of each working day, the information of unsatisfied customers is summarized, technicians are allocated to service tasks, and visiting routes are planned. The objective is to minimize the expected total discounted technicians’ travel and customers’ waiting costs. We propose an innovative hybrid value function approximation method, which is based on a graph neural network, that efficiently handles large decision spaces with candidate solutions and a hybrid genetic algorithm. Our approach outperforms two benchmarks in a comprehensive numerical study. We not only achieve a lower cost but also need less technician capacity. We illustrate that with our approach, a company can achieve large cost savings if customers accept a short waiting time.
