Skill generalization across different tasks is currently a challenging task for robots. As for recent works based on robot learning, substantial environmental interaction costs or abundant expert data are usually needed, thus causing great harm to the robot or the operating object. In this article, feature-selected adaptation transfer is proposed, aiming at accelerating the network learning process, and reducing the harm caused by the interaction process. Based on the domain adaptation, the source domain data with small maximum mean discrepancy to the target domain are extracted to pretrain the target domain policy. By extracting the shared features of the source domain and the target domain, the knowledge transfer between old task and new task is realized. Moreover, the data, more favorable to the target domain, are selected to update the network and further improve the stability of network training. Besides, a series of peg-in-hole tasks is conducted in simulation, and they can be transferred to the real world without directly interacting with the environment.

Given the problem of exploration and utilization, the high interaction cost of deep reinforcement learning is sometimes unacceptable. Moreover, the model trained by deep reinforcement learning cannot adapt to the complex and changeable assembly environment quickly owing to the weak generalization ability. Thus, the knowledge transfer framework is established for different robot peg-in-hole assemblies in this article to reinforce the generalization ability of the assembly model and the utilization efficiency of data. In the framework, the source domain model is trained through the Proximal Policy Optimization algorithm. The model accurately predicts environmental information and dynamically adjusts the robot’s movements based on the prediction. Combined with the PI controller, the flexible peg-in-hole assembly can be quickly achieved by the robot. The policy fusion between the source and target domain MDP is realized by the Policy Fusion Transfer algorithm designed in this article. The effectiveness of the algorithm is verified and discussed based on the transfer experiments, involving different assembly objects and different assembly environments. The training results demonstrate that the environment is explored by the Policy Fusion Transfer algorithm in a safe range at the initial stage of training by reusing the assembly policy of the source domain and that the faster model convergence speed is achieved. Furthermore, the testing results suggest that the assembly success rate is improved by 26% and the assembly force is reduced by 30N. Moreover, the knowledge transfer between different peg-in-hole assemblies is realized successfully.

Transfer learning and ensemble learning are currently emerging trends in robotics, playing a vital role in the generalization of robot skills for manipulation tasks. In the face of new tasks, previous methods either learn from scratch or require substantial data and experience specific to the target domain, significantly hindering the generalization of robot skills. In this article, an ensemble transfer learning method is proposed based on domain differences. First, a hybrid approach that combines sequential and parallel ensemble learning strategies is designed. The experience of strategies in different domains can be reused in the target domain through ensemble transfer. Second, a strategy weight adjustment method is designed by considering the differences among various domains. This allows the strategies of the source domains that are more similar to the target domain to be assigned greater weights, thus maximizing the utilization of the knowledge of the source domains. The ensemble strategy not only exhibits good generalization for new tasks but also has excellent memory for old tasks. This article combines sequential and parallel ensemble learning methods based on domain difference, which can quickly generalize strategies from multiple source domains to the target domain without learning in the target domain. The dominant base strategies are enhanced through weight adjustment, and the adverse impact of poor base strategies is avoided on the overall strategy. A series of simulation and real multiple peg-in-hole assembly experiments is conducted to verify the superiority of the proposed method.

Traditional assembly tasks often require robots to transfer the acquired skills to new tasks. However, previous transfer reinforcement learning methods typically ignore the inherent relationship between the source and the target domain tasks. This requires a substantial amount of interaction data to compensate for this deficiency, and generally results in poor transfer effects. To address this issue, a strategy transfer method that establishes a shared feature space between the source domain and the target domain is proposed to enhance the efficiency of strategy learning on peg-in-hole assembly. Initially, by calculating the distance between each feature in the source and target domains, the features with small distance are selected as shared features. Subsequently, in order to determine the successful search state, this paper uses the jump state of contact force and the relative position between the peg and the hole as the judgment criterion. Lastly, search and insertion peg-in-hole assembly experiments are conducted to validate the generalization of the proposed strategy, demonstrating its capability to transfer from simulation to the real world.

Cable-driven parallel robots (CDPRs) are increasingly favored in rehabilitation, medical devices, and material transportation due to their flexible structure and large transmission distance. The CDPRs with a highly modular and flexible structure are usually easy to be quickly reorganized. It is important to study the dimension and shape optimization of the basis and moving platforms for rapidly reconstructing a high-performance CDPR. The influence of each parameter of CDPRs’ dimension and shape on performance is mutually coupled. Therefore, obtaining the global optimal result by simply superimposing each optimum parameter is usually difficult. To this end, the concepts of a constant stiffness space (CSS) and a cable-tension-constrained workspace (CTCW) and their calculation methods are introduced, and the CDPRs’ dimension and shape are optimized with the maximum CSS and CTCW volume as the optimization indicators. First, the response surface optimization model between CDPRs’ performance and multi-objective optimization parameters is established, taking into account the coupling relationship of each CDPR optimization parameter and the effect on performance, and it is solved by using the Latin hypercube design method. Then, the effect of CDPRs’ dimension and shape on performance is analyzed by using the response surface optimization model, and the CDPRs’ optimization dimensions are provided.