Biomimetics

Flapping-Wing Aerial Vehicles (FWAVs), which take inspiration from the flight of birds and insects, have gained increasing attention over the past decades due to advantages such as low noise, biomimicry and safety, enabled by the absence of propellers. These features make them particularly suitable for applications in natural environments and operations near humans. However, their complexity introduces significant challenges, including difficulties in take-off and landing as well as limited endurance. Perching represents a promising solution to address these limitations. By equipping these drones with a perching mechanism, they could land on branches to save energy and later exploit the altitude to resume flight without requiring human intervention. Specifically, this review focuses on perching mechanisms based on grasping. It presents designs developed for flapping-wing platforms and complements them with systems originally intended for other types of aerial robots, evaluating their applicability to FWAV applications. The purpose of this work is to provide a structured overview of the existing strategies to support the development of new, effective solutions that could enhance the use of FWAVs in real-world applications. Full article

Aiming at the problems of high-dimensional nonlinear constraints, multi-objective conflicts, and low solution efficiency in microgrid optimal scheduling, this paper proposes a multi-objective Harris Hawk–Grey Wolf hybrid intelligent algorithm (IMOHHOGWO). The problem of balancing the global exploration and local exploitation of the algorithm is solved by introducing an adaptive energy factor and a nonlinear convergence factor; in terms of the algorithm’s exploration scope, the stochastic raid strategy of Harris Hawk optimization (HHO) is used to generate diversified solutions to expand the search scope, and constraints such as the energy storage SOC and DG outflow are finely tuned through the α/β/δ wolf bootstrapping of the Grey Wolf Optimizer (GWO). It is combined with a simulated annealing perturbation strategy to enhance the adaptability of complex constraints and local search accuracy, at the same time considering various constraints such as power generation, energy storage, power sales, and power purchase. We establish the microgrid multi-objective operation cost and carbon emission cost objective function, and through the simulation examples, we verify and determine that the IMOHHOGWO hybrid intelligent algorithm is better than the other three algorithms in terms of both convergence speed and convergence accuracy. According to the results of the multi-objective test function analysis, its performance is superior to the other four algorithms. The IMOHHOGWO hybrid intelligent algorithm reduces the grid operation cost and carbon emissions in the microgrid optimal scheduling model and is more suitable for the microgrid multi-objective model, which provides a feasible reference for future integrated microgrid optimal scheduling. Full article

To address the challenge of coordinating economic cost control and low-carbon objectives in microgrid scheduling, while overcoming the performance limitations of the traditional Zebra Optimization Algorithm (ZOA) in complex problems, this paper proposes a Synergistic Zebra Optimization Algorithm (SZOA) and integrates it with innovative management concepts to enhance the microgrid scheduling process. The SZOA incorporates three core strategies: a multi-population cooperative search mechanism to strengthen global exploration, a vertical crossover–mutation strategy to meet high-dimensional scheduling requirements, and a leader-guided boundary control strategy to ensure variable feasibility. These strategies not only improve algorithmic performance but also provide technical support for innovative management in microgrid scheduling. Extensive experiments on the CEC2017 (d = 30) and CEC2022 (d = 10, 20) benchmark sets demonstrate that the SZOA achieves higher optimization accuracy and stability compared with those of nine state-of-the-art algorithms, including IAGWO and EWOA. Friedman tests further confirm its superiority, with the best average rankings of 1.20 for CEC2017 and 1.08/1.25 for CEC2022 (d = 10, 20). To validate practical applicability, the SZOA is applied to grid-connected microgrid scheduling, where the system model integrates renewable energy sources such as photovoltaic (PV) generation and wind turbines (WT); controllable sources including fuel cells (FC), microturbines (MT), and gas engines (GS); a battery (BT) storage unit; and the main grid. The optimization problem is formulated as a bi-objective model minimizing both economic costs—including fuel, operation, pollutant treatment, main-grid interactions, and imbalance penalties—and carbon emissions, subject to constraints on generation limits and storage state-of-charge safety ranges. Simulation results based on typical daily data from Guangdong, China, show that the optimized microgrid achieves a minimum operating cost of USD 5165.96, an average cost of USD 6853.07, and a standard deviation of only USD 448.53, consistently outperforming all comparison algorithms across economic indicators. Meanwhile, the SZOA dynamically coordinates power outputs: during the daytime, it maximizes PV utilization (with peak output near 35 kW) and WT contribution (30–40 kW), while reducing reliance on fossil-based units such as FC and MT; at night, BT discharges (−20 to −30 kW) to cover load deficits, thereby lowering fossil fuel consumption and pollutant emissions. Overall, the SZOA effectively realizes the synergy of “economic efficiency and low-carbon operation”, offering a reliable and practical technical solution for innovative management and sustainable operation of microgrid scheduling. Full article

We consider optimization for different production requirements from the viewpoint of a bio-inspired framework for system flexibility that allows us to study the ability of an algorithm to transfer solutions from previous optimization tasks, which also relates to dynamic evolutionary optimization. Optimizing manufacturing process parameters is typically a multi-objective problem with often contradictory objectives, such as production quality and production time. If production requirements change, process parameters have to be optimized again. Since optimization usually requires costly simulations based on, for example, the Finite Element method, it is of great interest to have a means to reduce the number of evaluations needed for optimization. Based on the extended Oxley model for orthogonal metal cutting, we introduce a multi-objective optimization benchmark where different materials define related optimization tasks. We use the benchmark to study the flexibility of NSGA-II, which we extend by developing two variants: (1) varying goals, which optimizes solutions for two tasks simultaneously to obtain in-between source solutions expected to be more adaptable, and (2) active–inactive genotype, which accommodates different possibilities that can be activated or deactivated. Results show that adaption with standard NSGA-II greatly reduces the number of evaluations required for optimization for a target goal. The proposed variants further improve the adaption costs, where on average, the computational effort is more than halved in comparison to the non-adapted baseline. We note that further work is needed for making the methods advantageous for real applications. Full article

Osseodensification (OD) has emerged as a favorable osteotomy preparation technique that preserves and compacts autogenous bone along the osteotomy walls during site preparation, enhancing primary stability and implant osseointegration. While OD has demonstrated promising results in low-density trabecular bone, especially when used in conjunction with acid-etched (AE) implant surfaces, its efficacy in high-density cortical bone remains unclear—particularly in the context of varying implant surface characteristics. In this study, Grade V titanium alloy implants (Ti-6Al-4V, 4 mm × 10 mm) with deep threads, designated bone chambers and either as-machined (Mach) or AE surfaces were placed in 3.8 mm diameter osteotomies in the submandibular region of 16 adult sheep using either OD or conventional (Reg) drilling protocols. Insertion torque values (N·cm) were measured at the time of implant placement to evaluate primary stability. Mandibles were harvested at 3-, 6-, 12-, or 24-weeks post-implantation (n = 4 sheep/time point), and histologic sections were analyzed to quantify bone-to-implant contact (BIC) and bone area fractional occupancy (BAFO). Qualitative histological analysis confirmed successful osseointegration among all groups at each of the healing time points. No statistically significant differences were observed between OD and conventional drilling techniques in insertion torque (p > 0.628), BIC (p > 0.135), or BAFO (p > 0.060) values, regardless of implant surface type or healing interval. The findings indicate that neither drilling technique nor implant surface treatment significantly influences osseointegration in high density cortical bone. Furthermore, as the osteotomy was not considerably undersized, the use of OD instrumentation showed no signs of necrosis, inflammation, microfractures, or impaired osseointegration in dense cortical bone. Both OD and Reg techniques appear to be suitable for implant placement in dense bone, allowing flexibility based on surgeon preference and clinical circumstances. Full article

Forests are critical ecosystems that play a fundamental role in supporting biodiversity and maintaining climate stability. However, forest monitoring and exploration present huge challenges due to the vast scale and complex terrain. This paper proposes a novel bionic robot, PteroBot, designed to support a new paradigm for forest exploration inspired by the locomotion of Pteromyini. PteroBot is capable of regulating its gliding posture via a flexible membrane, enabling low-energy and low-disturbance mobility within forest environments. An adaptive gliding control system tailored to the robot’s structure is developed and its effectiveness is validated through aerodynamic analysis, simulation, and experimental testing. Results show that under a cascaded closed-loop attitude controller, PteroBot achieves an average glide ratio of 2.02 and demonstrates controllable turning via attitude modulation. Additionally, comparative tests with UAVs demonstrate that PteroBot offers significant advantages in energy efficiency and acoustic disturbance. Experimental outcomes confirm that PteroBot offers a biologically inspired and ecologically compatible solution for forest exploration, with strong potential in applications such as environmental monitoring, habitat assessment, and covert reconnaissance. Full article

The Zebra Optimization Algorithm (ZOA) is a metaheuristic algorithm inspired by the collective behavior of zebras in the wild. Like many other swarm intelligence algorithms, the ZOA faces several limitations, including slow convergence, susceptibility to local optima, and an imbalance between exploration and exploitation. To address these challenges, this paper proposes an improved version of the ZOA, termed the Multi-strategy Improved Zebra Optimization Algorithm (MIZOA). First, a multi-population search strategy is introduced to replace the traditional single population structure, dividing the population into multiple subpopulations to enhance diversity and improve global convergence. Second, the mutation operation of genetic algorithm (GA) is integrated with the Metropolis criterion to boost exploration capability in the early stages while maintaining strong exploitation performance in the later stages. Third, a novel selective aggregation strategy is proposed, incorporating the hunting behavior of the Coati Optimization Algorithm (COA) and Lévy flight to further enhance global exploration and convergence accuracy during the defense phase. Experimental evaluations are conducted on 23 benchmark functions, comparing the MIZOA with eight existing swarm intelligence algorithms. The performance is assessed using non-parametric statistical tests, including the Wilcoxon rank-sum test and the Friedman test. The results demonstrate that the MIZOA achieves superior global convergence accuracy and optimization performance, confirming its robustness and effectiveness. The MIZOA was evaluated on real-world engineering problems against seven algorithms to validate its practical performance. Furthermore, when applied to path planning tasks for Automated Guided Vehicles (AGVs), the MIZOA consistently identifies paths closer to the global optimum in both simple and complex environments, thereby further validating the effectiveness of the proposed improvements. Full article

In this study, a novel compliant ankle structure with three passive degrees of freedom for quadruped robots is presented. First, this paper introduced the bionic principle and structural implementation method of the passively compliant ankle, with a particular focus on the configuration and working principle of the elastic adjustment element. Then, the kinematic model of the ankle and mathematic model of the elastic element, comprising mechanical and pneumatic model, was established by using appropriate theory. Finally, a test rig of the ankle was carried out to verify its actual function. The research results show that: (1) The ankle structure demonstrates excellent stability, maintaining its upright posture even under unreliable foot–ground interactions. (2) Compared to traditional structure, the single-leg module incorporating the proposed design exhibits smoother forward stepping under an appropriate pre-inflation pressure, with its actual motion trajectory showing closer agreement with the planned one; (3) The parallel topology enables a notable reduction in the driving torque of each joint in the leg during motion, thereby improving the energy efficiency of robots. Full article

Online peer support platforms are vital, scalable resources for mental health, yet their effectiveness is frequently undermined by inefficient user matching, severe participation inequality, and subsequent “super-helper” burnout. This study introduces TrailMap, a novel peer-matching algorithm inspired by the decentralised foraging strategies of ant colonies. By treating user interactions as paths that gain or lose “pheromone” based on helpfulness ratings, the system enables the community to collectively and adaptively identify its most effective helpers. A two-phase validation study was conducted. First, an agent-based simulation demonstrated that TrailMap reduced the mean time to a helpful response by over 70% and improved workload equity compared to random routing. Second, a four-week randomised controlled pilot study with human participants confirmed these gains, showing a 76% reduction in median wait time and significantly higher perceived helpfulness ratings. The findings suggest that by balancing the workload, TrailMap enhances not only the efficiency but also the socio-technical sustainability of online support communities. TrailMap provides a practical, nature-inspired method for building more resilient and equitable online support communities, enhancing access to effective mental health support. Full article

In this study, we propose an architectural model for path optimization in cluster order picking within warehouse robotics, utilizing a hybrid approach that combines symbolic control and metaheuristic techniques. Among the optimization strategies, we incorporate bio-inspired metaheuristic algorithms such as the Walrus Optimization Algorithm (WOA), Puma Optimization Algorithm (POA), and Flying Foxes Algorithm (FFA), which are grounded in behavioral models observed in nature. We consider large-scale warehouse robotic systems, partitioned into clusters. To manage shared resources between clusters, the set of clusters is first formulated as a symbolic control design task within a discrete synthesis framework. Subsequently, the desired control goals are integrated into the model, encoded using parallel synchronous dataflow languages; the resulting controller, derived using our safety-focused and optimization-based synthesis approach, serves as the manager for the cluster. Safety objectives address the rigid system behaviors, while optimization objectives focus on minimizing the traveled path of the warehouse robots through the constructed cost function. The metaheuristic algorithms contribute at this stage, drawing inspiration from real-world animal behaviors, such as walruses’ cooperative movement and foraging, pumas’ territorial hunting strategies, and flying foxes’ echolocation-based navigation. These nature-inspired processes allow for effective solution space exploration and contribute to improving the quality of cluster-level path optimization. Our hybrid approach, integrating symbolic control and metaheuristic techniques, demonstrates significantly higher performance advantage over existing solutions, with experimental data verifying the practical effectiveness of our approach. Our proposed algorithm achieves up to 3.01% shorter intra-cluster paths compared to the metaheuristic algorithms, with an average improvement of 1.2%. For the entire warehouse, it provides up to 2.05% shorter paths on average, and even in the worst case, outperforms competing metaheuristic methods by 0.28%, demonstrating its consistent effectiveness in path optimization. Full article

In manufacturing settings where humans and machines collaborate, understanding and predicting human intention is crucial for enabling the seamless execution of tasks. This knowledge is the basis for creating an intelligent, symbiotic, and collaborative environment. However, current foundation models often fall short in directly anticipating complex tasks and producing contextually appropriate motion. This paper proposes a modular framework that investigates strategies for structuring task knowledge and engineering context-rich prompts to guide Vision–Language Models in understanding and predicting human intention in semi-structured environments. Our evaluation, conducted across three use cases of varying complexity, reveals a critical tradeoff between prediction accuracy and latency. We demonstrate that a Rolling Context Window strategy, which uses a history of frames and the previously predicted state, achieves a strong balance of performance and efficiency. This approach significantly outperforms single-image inputs and computationally expensive in-context learning methods. Furthermore, incorporating egocentric video views yields a substantial 10.7% performance increase in complex tasks. For short-term motion forecasting, we show that the accuracy of joint position estimates is enhanced by using historical pose, gaze data, and in-context examples. Full article

This study proposes an Improved Whale Migration Algorithm (IWMA) to overcome the shortcomings of the original Whale Migration Algorithm, which suffers from premature convergence and insufficient local exploitation in high-dimensional multimodal optimization. IWMA introduces three enhancements: circle chaotic initialization to improve population diversity, a three-layer cooperative search framework to achieve a stronger balance between exploration and exploitation, and a dynamic adaptive mechanism with t-distribution re-exploration to reinforce both global escaping and local refinement. On the CEC2017 benchmark suite, IWMA demonstrates clear superiority over seven representative algorithms, delivering the best results on 27 out of 29 functions by best, 25 by mean, and 23 by standard deviation in 30 dimensions, and on 25, 18, and 18 functions, respectively, in 50 dimensions. Compared with other migration-based optimizers, its average rank improves by more than 30 percent, while runtime analysis shows only a small additional overhead of 7 to 12 percent. These outcomes, supported by convergence curves, boxplots, radar charts, and Wilcoxon tests, confirm the effectiveness of the proposed improvements. In six multi-UAV path planning scenarios, IWMA reduces the average cost by 14.5 percent compared with WMA and achieves up to 32.1 percent reduction in the most complex case. Overall, its average cost decreases by 27.4 percent across seven competitors, with a 23.6 percent improvement in the best solutions. These results demonstrate that the proposed modifications are effective, enabling IWMA to transfer its performance gains from benchmark tests to practical multi-UAV cooperative mission planning, where it consistently produces safer and smoother trajectories under complex constraints. Full article

The control of a biped robot is a challenging task due to the hard-to-stabilize dynamics. Generating a suitable walking reference trajectory is a key aspect of this problem. This article proposes a novel method of generating walking patterns for biped robots. The method integrates the double support phase and the single support phase into one step, and uses this step as the unit for trajectory generation through quadratic optimization with terminal constraints based on the Linear Inverted Pendulum Model, enabling us to shorten the optimization horizon while still generating natural walking trajectories. Moreover, by restricting the position and acceleration of the center of mass (COM) in the vertical direction, an excessive constraint is formed on the Zero Moment Point (ZMP) to offset the nonlinear effects of the COM’s vertical motion on the ZMP. This allows the COM of the robot to change in the vertical direction while maintaining the linearity of the optimization problem. Finally, the performance of the proposed method is validated by simulations and experiments of walking on flat ground and stairs using a position-controlled biped robot, Neubot. Full article

Binarizing continuous metaheuristics to solve challenging NP-hard binary optimization problems is a fundamental step in adapting continuous algorithms for discrete domains. Binary optimization problems, such as the Set Covering Problem and the 0–1 Knapsack Problem, demand tailored approaches to efficiently explore and exploit the solution space. The process of binarization often introduces complexities, as it requires balancing the transformation of continuous populations into binary solutions while preserving the algorithm’s capability to navigate the search space effectively. In this context, we explore the performance of the Reptile Search Algorithm (RSA), a continuous metaheuristic, applied to these two benchmark problems. To address the binary nature of the problems, a two-step binarization process is implemented, utilizing combinations of transfer functions with binarization rules. This framework enables the RSA to generate binary solutions while leveraging its inherent strengths in exploration and exploitation. Comparative experiments are conducted with Particle Swarm Optimization and the Grey Wolf Optimizer to benchmark the RSA’s performance under similar conditions. These experiments analyze critical factors such as fitness values, convergence behavior, and exploration–exploitation dynamics, providing insights into the effectiveness of different binarization approaches. The results demonstrate that the RSA achieves competitive performance across both problems, highlighting its flexibility and adaptability, which are attributed to its diverse movement equations. Notably, the Z4 transfer function consistently enhances performance for all algorithms, even when paired with less effective binarization rules. This indicates the potential of Z4 as a robust transfer function for binary optimization. The findings underscore the importance of selecting appropriate binarization strategies to maximize the performance of continuous metaheuristics in binary domains, paving the way for further advancements in hybrid optimization methodologies. Full article

Hand rehabilitation exoskeletons play a critical role in restoring motor function in patients with stroke or hand injuries. However, most existing designs rely on fixed-axis assumptions, neglecting the rolling–sliding coupling of finger joints that causes instantaneous center of rotation (ICOR) drift, leading to kinematic misalignment and localized pressure concentrations. This study proposes the Instant Radius Method (IRM) based on medical imaging to continuously model ICOR trajectories of the MCP, PIP, and DIP joints, followed by the construction of an equivalent ICOR through curve fitting. Crossing-type biomimetic kinematic pairs were designed according to the equivalent ICOR and integrated into a three-loop ten-linkage exoskeleton capable of dual DOFs per finger (flexion–extension and abduction–adduction, 10 DOFs in total). Kinematic validation was performed using IMU sensors (Delsys) to capture joint angles, and interface pressure distribution at MCP and PIP was measured using thin-film pressure sensors. Experimental results demonstrated that with biomimetic kinematic pairs, the exoskeleton’s fingertip trajectories matched physiological trajectories more closely, with significantly reduced RMSE. Pressure measurements showed a reduction of approximately 15–25% in mean pressure and 20–30% in peak pressure at MCP and PIP, with more uniform distributions. The integrated framework of IRM-based modeling–equivalent ICOR–biomimetic kinematic pairs–multi-DOF exoskeleton design effectively enhanced kinematic alignment and human–machine compatibility. This work highlights the importance and feasibility of ICOR alignment in rehabilitation robotics and provides a promising pathway toward personalized rehabilitation and clinical translation. Full article

The accuracy of numerical model outputs strongly depends on the quality of the initial wind field, yet ground observation data are typically sparse and provide incomplete spatial coverage. More importantly, many current mainstream correction models rely on reanalysis grid datasets like ERA5 as the true value, which relies on interpolation calculation, which directly affects the accuracy of the correction results. To address these issues, we propose a new deep learning model, PPWNet. The model directly uses sparse and discretely distributed observation data as the true value, which integrates observation point positions and a physical consistency term to achieve a high-precision corrected wind field. The model design is inspired by biological intelligence. First, observation point positions are encoded as input and observation values are included in the loss function. Second, a parallel dual-branch DenseInception network is employed to extract multi-scale grid features, simulating the hierarchical processing of the biological visual system. Meanwhile, PPWNet references the PointNet architecture and introduces an attention mechanism to efficiently extract features from sparse and irregular observation positions. This mechanism reflects the selective focus of cognitive functions. Furthermore, this paper incorporates physical knowledge into the model optimization process by adding a learned physical consistency term to the loss function, ensuring that the corrected results not only approximate the observations but also adhere to physical laws. Finally, hyperparameters are automatically tuned using the Bayesian TPE algorithm. Experiments demonstrate that PPWNet outperforms both traditional and existing deep learning methods. It reduces the MAE by 38.65% and the RMSE by 28.93%. The corrected wind field shows better agreement with observations in both wind speed and direction, confirming the effectiveness of incorporating position information and a physics-informed approach into deep learning-based wind field correction. Full article

This study aimed to evaluate how different insertion angles of titanium orthodontic miniscrews (30°, 45°, and 90°) influence stress distribution and displacement in surrounding alveolar bone using three-dimensional finite element analysis (FEA), with a focus on biomechanical outcomes at the titanium–bone interface. The 90° insertion angle generated the highest stress in cortical bone (58.2 MPa) but the lowest displacement (0.023 mm), while the 30° angle produced lower stress (36.4 MPa) but greater displacement (0.052 mm). The 45° angle represented a compromise, combining moderate stress (42.7 MPa) and displacement (0.035 mm). This simulation-based study was conducted between January and April 2025 at the Department of Orthodontics, Kocaeli Health and Technology University. A standardized 3D mandibular bone model (2 mm cortical and 13 mm cancellous layers) was constructed, and Ti-6Al-4V miniscrews (1.6 mm × 8 mm) were virtually inserted at 30°, 45°, and 90°. A horizontal orthodontic load of 2 N was applied, and von Mises stress and displacement values were calculated in ANSYS Workbench. Stress patterns were visualized using color-coded maps. The 90° insertion angle generated the highest von Mises stress in cortical bone (50.6 MPa), with a total maximum stress of 58.2 MPa, followed by 45° (42.7 MPa) and 30° (36.4 MPa) insertions (p < 0.001). Stress was predominantly concentrated at the cortical entry point, especially in the 90° model. In terms of displacement, the 90° group exhibited the lowest mean displacement (0.023 ± 0.002 mm), followed by 45° (0.035 ± 0.003 mm) and 30° (0.052 ± 0.004 mm), with statistically significant differences among all groups (p < 0.001). The 45° angle showed a balanced biomechanical profile, combining moderate stress and displacement values, as confirmed by post hoc analysis. From a biomimetics perspective, understanding how insertion angle affects bone response provides insights for designing bio-inspired anchorage systems. By simulating natural stress dissipation, this study demonstrates that insertion angle strongly modulates miniscrew performance. Vertical placement (90°) ensures rigidity but concentrates cortical stress, whereas oblique placement, particularly at 45°, offers a balanced compromise with adequate stability and reduced stress. These results emphasize that beyond material properties, surgical parameters such as insertion angle are critical for clinical success. Full article

Backward running is common in handball defense and relies heavily on proprioceptive control when visual information is limited. Twenty-eight male handball players were allocated to three groups: experimental novice group with blindfold training (n = 7), control novice group with the same training without blindfold (n = 7), and target professional group (n = 14). Both novice groups completed a 6-week balance program (3 × 20 min/week). Lower-limb kinematics during backward running were captured with a 6-camera motion analysis system, and inter-joint coordination was quantified by Mean Absolute Relative Phase (MARP) and Deviation Phase (DP) for ankle–knee and knee–pelvic couplings. At baseline, professionals showed greater ankle–knee MARP than novices (ANOVA F(2,25) = 9.42, p < 0.001). Representative means (mean ± SD): ankle–knee MARP novices 1.62–1.79 vs. professionals 3.83. After training, ankle–knee MARP increased in both novice groups (experimental: t(6) = 4.72, p < 0.001; control: t(6) = 5.02, p < 0.001), approaching professional values (post-training novices ≈ 3.22–3.26). Post-training between-group differences were non-significant for ankle–knee MARP (ANOVA F(2,25) = 1.24, p = 0.30), while ankle–knee DP showed a group effect (F(2,25) = 5.12, p = 0.01; experimental vs. professional t(19) = 3.12, p = 0.01). A short-term balance program improved ankle–knee coordination during backward running in novice male players; additional blindfolding did not yield extra benefit over 6 weeks. These findings can inform short-term training and rehabilitation planning for handball, while long-term effects require future study. Full article

This study addresses the core challenges of precise marigold corollas recognition and efficient continuous path planning under complex natural conditions (strong illumination, occlusion, adhesion) by proposing an integrated lightweight visual recognition and real-time path planning framework. We introduce MPD-YOLO, an optimized model based on YOLOv11n, incorporating (1) a Multi-scale Information Enhancement Module (MSEE) to boost feature extraction; (2) structured pruning for significant model compression (final size: 2.1 MB, 39.6% of original); and (3) knowledge distillation to recover accuracy loss post-pruning. The resulting model achieves high precision (P: 89.8%, mAP@0.5: 95.1%) with reduced computational load (3.2 GFLOPs) while demonstrating enhanced robustness in challenging scenarios—recall significantly increased by 6.8% versus YOLOv11n. Leveraging these recognition outputs, an adaptive ant colony algorithm featuring dynamic parameter adjustment and an improved pheromone strategy reduces average path planning time to 2.2 s—a 68.6% speedup over benchmark methods. This integrated approach significantly enhances perception accuracy and operational efficiency for automated marigold harvesting in unstructured environments, providing robust technical support for continuous automated operations. Full article