RipenessGAN is a novel Generative Adversarial Network (GAN) designed to generate synthetic images across different ripeness stages of jujubes (green fruit, white ripe fruit, semi-red fruit, and fully red fruit), aiming to provide balanced training data for diverse applications beyond classification accuracy. This study addresses the problem of data imbalance by augmenting each ripeness stage using our proposed Growth Day Embedding mechanism, thereby enhancing the performance of downstream classification models. The core innovation of RipenessGAN lies in its ability to capture continuous temporal transitions among discrete ripeness classes by incorporating fine-grained growth day information (0–56 days) in addition to traditional class labels. The experimental results show that RipenessGAN produces synthetic data with higher visual quality and greater diversity compared to CycleGAN. Furthermore, the classification models trained on the enriched dataset exhibit more consistent and accurate performance. We also conducted comprehensive comparisons of RipenessGAN against CycleGAN and class-conditional diffusion models (DDPM) under strictly controlled and fair experimental settings, carefully matching model architectures, computational resources, training conditions, and evaluation metrics. The results indicate that although diffusion models yield highly realistic images and CycleGAN ensures stable cycle-consistent generation, RipenessGAN provides superior practical benefits in training efficiency, temporal controllability, and adaptability for agricultural applications. This research demonstrates the potential of RipenessGAN to mitigate data imbalance in agriculture and highlights its scalability to other crops.

Motion-capture data provide high accuracy but are difficult to obtain, necessitating dataset augmentation. To our knowledge, no prior study has investigated few-shot generative models for motion-capture data that address both quality and diversity. We tackle the diversity loss that arises with extremely small datasets (n ≤ 10) by applying transfer learning and continual learning to retain the rich variability of a larger pretraining corpus. To assess quality, we introduce MFMMD (Motion Feature-Based Maximum Mean Discrepancy)—a metric well-suited for small samples—and evaluate diversity with the multimodality metric. Our method embeds an Elastic Weight Consolidation (EWC)-based regularization term in the generator’s loss and then fine-tunes the limited motion-capture set. We analyze how the strength of this term influences diversity and uncovers motion-specific characteristics, revealing behavior that differs from that observed in image-generation tasks. The experiments indicate that the transfer learning pipeline improves generative performance in low-data scenarios. Increasing the weight of the regularization term yields higher diversity in the synthesized motions, demonstrating a marked uplift in motion diversity. These findings suggest that the proposed approach can effectively augment small motion-capture datasets with greater variety, a capability expected to benefit applications that rely on diverse human-motion data across modern robotics, animation, and virtual reality.

Motion capture data is crucial but creating a large dataset can be challenging due to complexities in acquisition. Generative Adversarial Network (GAN)-based motion data augmentation offers a potential solution to this issue. However, GANs often struggle with learning from limited data, resulting in poor quality output. In this study, we propose a Dynamic Time Warping (DTW) filtering method that filters out generated data significantly deviating from real-world examples. Through this approach, we have achieved an improvement in the fidelity of the generated data, even with dataset size constraints, as evidenced by an increase in action recognition accuracy.

Multi-robot task allocation can place relay and mission goals in one matrix; whether this controls geometric relay-chain feasibility in execution had not been established. A deployed unified formulation was evaluated in closed-loop Robot Operating System 2 (ROS 2)/Gazebo simulation on six identical robots with synthetic ranges and no simulated radio, as component-level assurance for human-supervised systems, not a mission-success estimate. As deployed, a mission task displaced a relay in 20/20 trials; a cost-divisor capability model assigned hops beyond holders’ ranges. After two established repairs, capability-conditioned relay placement (CCRP) chose relay positions and holders jointly, bounding each hop by both holders’ ranges. On the mixed-range fleet, CCRP formed a geometrically feasible chain in 10/10 trials, in $16.5$ s (median) against $21.3$ s for equal-hop placement in one session (in-session gains $1.1$–$10.05$ s, not all significant). Across five arrangement–target cells, plans were reproduced in 100/100 trials. CCRP’s median formation time was lower in all four cells whose holders differed, significantly in two; in one, the capability-blind fall-through held no chain. CCRP produced geometrically feasible plans and, on the mixed-range fleet, faster chain formation; stalls beside stationary robots remained outside its distance objective. No transfer to untested maps, radios or hardware is claimed.