It is very challenging to precisely control a redundant manipulator with an unknown model during the end-effector tracking task. Adata-driven approach offers a promising solution for manipulator control under such conditions. In this article, two data-driven discrete neurodynamics (DDDN) schemes are proposed for redundant manipulator tracking control. First, utilizing discrete neurodynamics (DN) principles, the DDDN-1 scheme with an adaptive Jacobian matrix is developed. Subsequently, the DDDN-2 scheme is further presented, which eliminates the need for the Jacobian matrix inversion operation. Detailed theoretical analyses verify the effectiveness of DDDN-1 and DDDN-2 schemes. Additionally, detailed comparisons with existing schemes have been provided. Finally, simulative and physical experiments conducted using the UR5 manipulator validate the theoretical analyses, demonstrating the effectiveness and superiority of DDDN-1 and DDDN-2 schemes.

Repetitive motion control of redundant manipulators typically requires precise kinematic models to construct Jacobian matrices. However, model-based approaches are inherently limited when manipulator parameters are unavailable or only partially known. This paper introduces a novel data-driven discrete zeroing neurodynamics (DDZN) model for repetitive motion control. Specifically, a Jacobian matrix estimation method based on data-driven technology is proposed, which eliminates the need for prior models by leveraging historical input-output information. By integrating the Jacobian matrix estimation with a discrete zeroing neurodynamics (DZN) model, the approach enables simultaneous trajectory tracking and repeatable configuration recovery without relying on structural parameters. Theoretical analysis verifies the performance of DDZN model under noise environment. Furthermore, abundant experiment results validate its reliability and superior performance compared with various models.

Meta-Black-Box Optimization (MetaBBO) streamlines the automation of optimization algorithm design through meta-learning. It typically employs a bi-level structure: the meta-level policy undergoes meta-training to reduce the manual effort required in developing algorithms for low-level optimization tasks. The original MetaBox (2023) provided the first open-source framework for reinforcement learning-based single-objective MetaBBO. However, its relatively narrow scope no longer keep pace with the swift advancement in this field. In this paper, we introduce MetaBox-v2 (\url{https://github.com/MetaEvo/MetaBox}) as a milestone upgrade with four novel features: 1) a unified architecture supporting RL, evolutionary, and gradient-based approaches, by which we reproduce $23$ up-to-date baselines; 2) efficient parallelization schemes, which reduce the training/testing time by $10-40$x; 3) a comprehensive benchmark suite of $18$ synthetic/realistic tasks ($1900$+ instances) spanning single-objective, multi-objective, multi-model, and multi-task optimization scenarios; 4) plentiful and extensible interfaces for custom analysis/visualization and integrating to external optimization tools/benchmarks. To show the utility of MetaBox-v2, we carry out a systematic case study that evaluates the built-in baselines in terms of the optimization performance, generalization ability and learning efficiency. Valuable insights are concluded from thorough and detailed analysis for practitioners and those new to the field.