Search Results (19)

This paper presents a large-scale behavioural-cloning benchmark of linear-programming (LP) truck dispatch in commercial open-pit mining, quantifying how much of the LP policy is recoverable from observable cycle records and which learning models recover it. Drawing on 983,025 LP dispatch decisions across four operational years at a large copper mine, four learned model families are compared under an identical, strictly causal feature set and a strict temporal hold-out (Year 5)—a 72,681-parameter multilayer perceptron (MLP), Random Forest, LightGBM, and XGBoost—against five non-parametric baselines. Gradient-boosted trees recover substantially more of the LP policy than the MLP: XGBoost attains 41.04% top-one and 79.50% top-three accuracy (95% CI [40.80, 41.30]), and LightGBM 39.70%/77.95%, both significantly exceeding the cycle-continuity heuristic (35.21%/67.32%) and the MLP (32.4%/68.8%) by McNemar tests (all p < 0.001). The dataset is highly imbalanced (normalized entropy 0.814; imbalance ratio 16,921:1), and front-end-loader classes with negligible support are not learnable. Permutation analysis shows the truck’s previous shovel dominates the MLP policy (+7.46 pp), yet XGBoost exceeds the previous-shovel-only Bayes-optimal accuracy of 32.40%, demonstrating that observable features beyond previous shovel carry exploitable signal the MLP fails to capture. A learning-curve ablation shows the gradient-boosting advantage is attributable to model architecture rather than training-data volume and is robust to hyperparameter choice, consistent with the established behaviour of tree ensembles on tabular data. The results indicate a learnability ceiling that sits well above the MLP and is partly model-limited rather than purely informational; they also show that imitation fidelity is distinct from dispatch quality, which is not assessed here. The study reframes behavioural cloning of commercial FMS dispatch as a diagnostic and benchmarking tool and motivates model choice, imbalance-aware learning, and richer state recovery as the levers for data-driven dispatch analysis. Full article

This paper studies large-scale last-mile delivery using a heterogeneous fleet of trucks, onboard drones in a hybrid truck–drone mode, and independent drones. Orders are first screened by a feasibility check; feasible orders are then assigned to one of the three modes by a delivery mode selection policy and routed using mode-specific planning algorithms. The delivery mode selection policy is trained with Proximal Policy Optimization (PPO), warm-started by behaviour cloning from heuristic decisions. For route planning, we use a five-step procedure for the hybrid mode and simple depot round trips for independent drones. Experiments on Solomon VRPTW benchmarks and extended instances (100/200/400 customers; R/C/RC distributions) show lower total cost than representative heuristic baselines and metaheuristics, with practical runtime. Sensitivity analysis over fleet sizes further indicates competitive performance across a range of truck and drone configurations, especially for medium and large fleets. Full article

Reinforcement learning (RL) represents an effective approach for developing autonomous navigation and obstacle avoidance capabilities in hospital automated guided vehicles (AGVs). However, real-world adoption is challenged by the need for carefully designed reward functions, low sample efficiency, and slow convergence behaviour. To effectively address these issues, in this work, BEAGM-PPO, a reinforcement learning framework tailored for unknown hospital environments, was proposed. A reference model was initially employed to improve sample efficiency by directing the agent’s learning process. The reference model consists of expert demonstrations and policy derivation mechanisms. During the expert demonstration phase, human experts perform the required tasks and generate state-action pair datasets for training. During the policy derivation phase, demonstration data, behaviour cloning, and uncertainty estimation were used to derive the imitated expert policy. An ant colony optimization (ACO)-inspired pheromone mechanism and a memory replay strategy were incorporated to improve target-oriented action selection and supress unnecessary exploration. Experiments conducted in typical 3D simulation scenarios demonstrated that the proposed method achieved the highest arrival rate compared with baseline models. Moreover, the integrated imitation learning approach enables uncertainty estimation for both the policy and the model, while expanded training datasets further enhance performance. Overall, the results prove that BEAGM-PPO serves as a solid theoretical foundation for autonomous navigation in hospital AGVs. Full article

Megaplatypus mutatus, a major poplar pest in South America, tunnels into the xylem, weakening trunks and reducing wood quality. Volatile organic compounds (VOCs) are key mediators of plant–insect interactions and may reflect genotype-specific defence strategies. This study analysed VOC profiles of young and adult Populus deltoides cv. Harvard and P. × canadensis cv. Conti 12 under natural M. mutatus infestation. Gas chromatography–mass spectrometry putatively annotated 31 VOCs, including green leaf volatiles (GLVs), pentyl leaf volatiles (PLVs), terpenes, alcohols, aromatics and phenolics, 12 of which, to our knowledge, have not been previously reported in Populus VOC profiles. Harvard trees showed ~14.5-fold higher total VOC abundance than Conti trees. In Conti, constitutive VOC emissions remained stable regardless of infestation status or age. In contrast, under infestation, Harvard trees emitted10-fold higher constitutive VOCs than non-infested Harvard trees and ~52-fold higher than Conti, a pattern consistent with increased defensive activity. GLVs and PLVs relatively dominated both genotypes, although Harvard showed higher emissions. Terpenes were not detected in young Conti trees under our analytical conditions but were abundant and diverse in infested Harvard trees, which may indicate a stronger terpene-associated response in this clone. Several compounds were detected only under specific genotype–condition combinations in our dataset and therefore represent candidate volatiles for future behavioural and functional studies. These results are consistent with differences in VOC emission patterns between genotypes and age classes, improve our understanding of putative chemical cues in the interaction between Populus and M. mutatus, and provide a basis for future work towards sustainable pest management strategies. Full article

Despite rapid advances in robot learning, the coordination of closed-loop manipulation in cluttered environments remains a challenging and relatively underexplored problem. We present a novel two-level hierarchical architecture for a depth vision-equipped robotic arm that integrates pushing, grasping, and high-level decision making. Central to our approach is a prioritised action–selection mechanism that facilitates efficient early-stage learning via behaviour cloning (BC), while enabling scalable exploration through reinforcement learning (RL). A high-level decision neural network (DNN) selects between grasping and pushing actions, and two low-level action neural networks (ANNs) execute the selected primitive. The DNN is trained with RL, while the ANNs follow a hybrid learning scheme combining BC and RL. Notably, we introduce an automated demonstration generator based on oriented bounding boxes, eliminating the need for manual data collection and enabling precise, reproducible BC training signals. We evaluate our method on a challenging manipulation task involving five closely packed cubic objects. Our system achieves a completion rate (CR) of 100%, an average grasping success (AGS) of 93.1% per completion, and only 7.8 average decisions taken for completion (DTC). Comparative analysis against three baselines—a grasping-only policy, a fixed grasp-then-push sequence, and a cloned demonstration policy—highlights the necessity of dynamic decision making and the efficiency of our hierarchical design. In particular, the baselines yield lower AGS (86.6%) and higher DTC (10.6 and 11.4) scores, underscoring the advantages of content-aware, closed-loop control. These results demonstrate that our architecture supports robust, adaptive manipulation and scalable learning, offering a promising direction for autonomous skill coordination in complex environments. Full article

Afforestation and reforestation (A/R) of non-agricultural and marginal saline lands by promoting fast-growing and salinity-tolerant woody species are crucial strategies to overcome land degradation and vegetation cover scarcity. To obtain basic information before using Populus alba clones in such degraded areas, morpho-physiological and cellular responses to salt stress were investigated. The experiment was conducted in a nursery where cuttings of three P. alba clones (MA-104, MA-195 and OG) were grown for 90 days in 100 mM NaCl versus a non-saline control. A global approach highlighting clonal differences in terms of dry mass production and plant physiological performance was achieved by comparing plant water status, gas exchange, ionic selectivity, osmotic adjustment and chloroplast ultrastructure under the two treatments. Dry mass production and eco-physiological processes were reduced in response to salt stress, with substantial clonal variation. Clone MA-104 exhibited salinity-tolerant behaviour in contrast to clone MA-195 and OG’s medium or sensitive behaviour towards the stress. Tolerance mechanisms may be attributed to enhanced stomatal control and osmotic adjustment, thereby enabling the maintenance of turgor in plants subjected to salt stress. The chloroplast ultrastructure also showed modifications that are often involved in adaptation to salinity stress. Full article

This study investigates how the behavioural realism of avatars can enhance virtual reality (VR) social interactions involving self-disclosure. First, we review how factors such as trust, enjoyment, and nonverbal communication could be influenced by motion capture technology by enhancing behavioural realism. We also address a gap in the prior literature by comparing different motion capture systems and how these differences affect perceptions of realism, enjoyment, and eye contact. Specifically, this study compared two types of avatars: an iClone UNREAL avatar with full-body and facial motion capture and a Vive Sync avatar with limited motion capture for self-disclosure. Our participants rated the iClone UNREAL avatar higher for realism, enjoyment, and eye contact duration. However, as shown in our post-experiment survey, some participants reported that they preferred the avatar with less behavioural realism. We conclude that a higher level of behavioural realism achieved through more advanced motion capture can improve the experience of VR social interactions. We also conclude that despite the general advantages of higher motion capture, the simpler avatar was still acceptable and preferred by some participants. This has important implications for improving the accessibility of avatars for different contexts, such as therapy, where simpler avatars may be sufficient. Full article

Recently, prokaryotic laccases from lactic acid bacteria (LAB), which can degrade biogenic amines, were discovered. A laccase enzyme has been cloned from Oenococcus oeni, a very important LAB in winemaking, and it has been expressed in Escherichia coli. This enzyme has similar characteristics to those previously isolated from LAB as the ability to oxidize canonical substrates such as 2,2-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS), 2,6-dimethoxyphenol (2,6-DMP), and potassium ferrocyanide K₄[Fe(CN₆)], and non-conventional substrates as biogenic amines. However, it presents some distinctiveness, the most characteristic being its psychrophilic behaviour, not seen before among these enzymes. Psychrophilic enzymes capable of efficient catalysis at low temperatures are of great interest due to their potential applications in various biotechnological processes. In this study, we report the discovery and characterization of a new psychrophilic laccase, a multicopper oxidase (MCO), from the bacterium Oenococcus oeni. The psychrophilic laccase gene, designated as LcOe 229, was identified through the genomic analysis of O. oeni, a Gram-positive bacterium commonly found in wine fermentation. The gene was successfully cloned and heterologously expressed in Escherichia coli, and the recombinant enzyme was purified to homogeneity. Biochemical characterization of the psychrophilic laccase revealed its optimal activity at low temperatures, with a peak at 10 °C. To our knowledge, this is the lowest optimum temperature described so far for laccases. Furthermore, the psychrophilic laccase demonstrated remarkable stability and activity at low pH (optimum pH 2.5 for ABTS), suggesting its potential for diverse biotechnological applications. The kinetic properties of LcOe 229 were determined, revealing a high catalytic efficiency (kcat/Km) for several substrates at low temperatures. This exceptional cold adaptation of LcOe 229 indicates its potential as a biocatalyst in cold environments or applications requiring low-temperature processes. The crystal structure of the psychrophilic laccase was determined using X-ray crystallography demonstrating structural features similar to other LAB laccases, such as an extended N-terminal and an extended C-terminal end, with the latter containing a disulphide bond. Also, the structure shows two Met residues at the entrance of the T1Cu site, common in LAB laccases, which we suggest could be involved in substrate binding, thus expanding the substrate-binding pocket for laccases. A structural comparison of LcOe 229 with Antarctic laccases has not revealed specific features assigned to cold-active laccases versus mesophilic. Thus, further investigation of this psychrophilic laccase and its engineering could lead to enhanced cold-active enzymes with improved properties for future biotechnological applications. Overall, the discovery of this novel psychrophilic laccase from O. oeni expands our understanding of cold-adapted enzymes and presents new opportunities for their industrial applications in cold environments. Full article

Artificial Intelligence bots are extensively used in multiplayer First-Person Shooter (FPS) games. By using Machine Learning techniques, we can improve their performance and bring them to human skill levels. In this work, we focused on comparing and combining two Reinforcement Learning training architectures, Curriculum Learning and Behaviour Cloning, applied to an FPS developed in the Unity Engine. We have created four teams of three agents each: one team for Curriculum Learning, one for Behaviour Cloning, and another two for two different methods of combining Curriculum Learning and Behaviour Cloning. After completing the training, each agent was matched to battle against another agent of a different team until each pairing had five wins or ten time-outs. In the end, results showed that the agents trained with Curriculum Learning achieved better performance than the ones trained with Behaviour Cloning by a matter of 23.67% more average victories in one case. In terms of the combination attempts, not only did the agents trained with both devised methods had problems during training, but they also achieved insufficient results in the battle, with an average of 0 wins. Full article

This study employed a coarse-grained Monte Carlo (MC) simulation to investigate the radiation-induced polymerisation of methacrylic acid (MAA) in an aqueous solution. This method provides an alternative to traditional kinetic models, enabling a detailed examination of the micro-structure and growth patterns of MAA polymers, which are often not captured in other approaches. In this work, we generated multiple clones of a simulation box, each containing a specific chemical composition. In these simulations, every coarse-grained (CG) bead represents an entire monomer. The growth function, defined by the chemical behaviour of interacting substances, was determined through repeated random sampling. This approach allowed us to simulate the complex process of radiation-induced polymerisation, enhancing our understanding of the formation of poly(methacrylic acid) hydrogels at a microscopic level; while Monte Carlo simulations have been applied in various contexts of polymerisation, this study’s specific approach to modelling the radiation-induced polymerisation of MAA in an aqueous environment, utilising the data obtained by quantum chemistry modelling, with an emphasis on micro-structural growth, has not been extensively explored in existing studies. This understanding is important for advancing the synthesis of these hydrogels, which have potential applications in diverse fields such as materials science and medicine. Full article

The main objective in this study is to describe and offer an account of verbal non-local doubling in Patagonian Spanish (PatSp), an understudied non-standard variety of Spanish in Argentina. We focus on data in which there are duplicated verbs surrounding an XP that bears the nuclear accent of the phrase (XP^NA). First, our analysis describes the prosodic, semantic, and morphosyntactic behaviour of the data gathered. Second, we present the problems and challenges that doubling phenomena in PatSp pose for approaches that have tried to explain similar data in other Spanish varieties and other languages, such as the copy theory or prosodic cloning. Third, this work explores a biclausal analysis of verbal non-local doubling in PatSp in which each duplicate originates in a different clause, CP1 and CP2. In this approach, duplicated verbs (V1 and V2, according to their linear distribution) are not derivationally related. We also argue that the XP^NA moves to the left periphery of CP2. This movement would account for the three typical traits of verbal duplication in PatSp: the mandatory adjacency between the nuclear accent and V2, the non-locality between verbal duplicates, and the semantic value of mirativity. Full article

Cassava is an essential tuber crop used to produce food, feed, and beverages. Whitefly pests, including Aleurothrixus aepim (Goeldi, 1886) (Hemiptera: Aleyrodidae), significantly affect cassava-based agroecosystems. Plant odours have been described as potential pest management tools, and the cassava clone M Ecuador 72 has been used by breeders as an essential source of resistance. In this study, we analysed and compared the volatile compounds released by this resistant clone and a susceptible genotype, BRS Jari. Constitutive odours were collected from young plants and analysed using gas chromatography–mass spectrometry combined with chemometric tools. The resistant genotype released numerous compounds with previously described biological activity and substantial amounts of the monoterpene (E)-β-ocimene. Whiteflies showed non-preferential behaviour when exposed to volatiles from the resistant genotype but not the susceptible genotype. Furthermore, pure ocimene caused non-preferential behaviour in whiteflies, indicating a role for this compound in repellence. This report provides an example of the intraspecific variation in odour emissions from cassava plants alongside information on odorants that repel whiteflies; these data can be used to devise whitefly management strategies. A better understanding of the genetic variability in cassava odour constituents and emissions under field conditions may accelerate the development of more resistant cassava varieties. Full article

Image-to-point cloud registration refers to finding relative transformation between the camera and the reference frame of the 3D point cloud, which is critical for autonomous driving. Recently, a two-stage “frustum point cloud classification + camera pose optimization” pipeline has shown impressive results on this task. This paper focuses on the second stage and reformulates the optimization procedure as a Markov decision process. An initial pose is modified incrementally, sequentially aligning a virtual 3D point observation towards a previous classification solution. We consider such an iterative update process as a reinforcement learning task and, to this end, propose a novel agent (AgentI2P) to conduct decision making. To guide AgentI2P, we employ behaviour cloning (BC) and reinforcement learning (RL) techniques: cloning an expert to learn accurate pose movement and reinforcing an alignment reward to improve the policy further. [We demonstrate the effectiveness and efficiency of our approach on Oxford Robotcar and KITTI datasets. The (RTE, RRE) metrics are $(1.34\mathrm{m},1.{46}^{\circ })$ on Oxford Robotcar and $(3.90\mathrm{m},5.{94}^{\circ })$ on KITTI, and the inference time is 60 ms, both achieving state-of-the-art performance]. The source code will be publicly available upon publication of the paper. Full article

Solid tumours are characterised by an altered microenvironment (TME) from the physicochemical point of view, displaying a highly hypoxic and acidic interstitial fluid. Hypoxia results from uncontrolled proliferation, aberrant vascularization and altered cancer cell metabolism. Tumour cellular apparatus adapts to hypoxia by altering its metabolism and behaviour, increasing its migratory and metastatic abilities by the acquisition of a mesenchymal phenotype and selection of aggressive tumour cell clones. Extracellular acidosis is considered a cancer hallmark, acting as a driver of cancer aggressiveness by promoting tumour metastasis and chemoresistance via the selection of more aggressive cell phenotypes, although the underlying mechanism is still not clear. In this context, Ca²⁺ channels represent good target candidates due to their ability to integrate signals from the TME. Ca²⁺ channels are pH and hypoxia sensors and alterations in Ca²⁺ homeostasis in cancer progression and vascularization have been extensively reported. In the present review, we present an up-to-date and critical view on Ca²⁺ permeable ion channels, with a major focus on TRPs, SOCs and PIEZO channels, which are modulated by tumour hypoxia and acidosis, as well as the consequent role of the altered Ca²⁺ signals on cancer progression hallmarks. We believe that a deeper comprehension of the Ca²⁺ signalling and acidic pH/hypoxia interplay will break new ground for the discovery of alternative and attractive therapeutic targets. Full article

The Ship’s Digital Twin (SDT) is a digital record of a ship’s behaviour or a software clone, which can be used to simulate scenarios that are expensive or hardly feasible to perform on a real object and especially in real time. The purpose of the SDT is to achieve cost reduction, obtain timely warnings of irregularities, and optimise individual ship system performances or the operation of the whole ship and to assist ship management. The aim of this paper is to describe the concept of the SDT and clarify some perplexities that may occur from initial introduction to concept. To that end, the paper identifies the steps in the SDT formulation process and methods used in each step of the process. Furthermore, a four-step iterative procedure for the SDT development is proposed. The applications of the concept are numerous, and some of them are presented in a review analysis in this paper. The presented analysis leads to a conclusion that should give some direction to future research in this area. Full article