ANTISPUBLISHERANTISPUBLISHER

Journal of Artificial Intelligence and Digital EconomyJournal of Artificial Intelligence and Digital Economy

Flexible robots pose a great challenge when partial manipulation is to be performed under unstructured environments. This paper deals with the issue of specific endpoint control and vibration cancellation of a multi-degree-of-freedom (DOF) tendon-driven flexible arm. The main shortcoming of standard controllers like PID is that they cannot respond to the nonlinearities, time-varying dynamics, and external disturbances inherent in the system. The methodology consists of detailed mechanical design through CAD/FEA, kinematic and dynamic modulization through a piecewise constant curvature approximation and synthesis of an online adaptive fuzzy inference system. The proposed system is tested with the help of simulation and physical testing. The major findings show that the trajectory tracking error (RMSE) decreased by 55.6 %, the amplitude of vibrations reduced by 70 % and the grasping success rates increased significantly, 40 % (PID) to 75 % (AFLC) with a fine stress ball. The settling time and the overshoot are also reduced by 45.2% and 61.6% respectively to the baseline PID controller by the AFLC.

The suggested Adaptive Fuzzy Logic Control (AFLC) strategy is effective in managing the nonlinearities of the system and its time-dependent parameters, performing much better than a standard PID controller both in simulation and real-world grasping tasks.Its adaptive nonlinear mapping ability is its secret of high performance.The AFLC is also able to change its behavior dynamically in response to the nonlinearities and time-varying parameters.The resultant performances are evidence-based that the consolidated AFLC strategy is indeed effective in solving the main issues of operating a multi-DOF flexible arm.Future work is suggested to incorporate computer vision for autonomous object detection and grasp pose estimation, investigate deep reinforcement learning for controller adaptation, and miniaturize the design for broader applications.

Berdasarkan penelitian ini, beberapa saran penelitian lanjutan dapat diajukan. Pertama, integrasi sistem visi komputer dapat memungkinkan robot untuk secara otomatis mendeteksi objek dan menentukan pose cengkeraman yang optimal, sehingga meningkatkan otonomi dan fleksibilitas sistem. Kedua, eksplorasi metode pembelajaran penguatan mendalam dapat mengoptimalkan adaptasi parameter pengontrol fuzzy secara real-time, memungkinkan robot untuk beradaptasi dengan perubahan lingkungan dan tugas yang kompleks. Ketiga, miniaturisasi desain robot dan pengembangan aktuator berkinerja tinggi dapat memperluas aplikasi sistem ke bidang-bidang seperti robotika medis dan eksplorasi ruang angkasa, membuka peluang baru untuk inovasi dan pemecahan masalah.

  1. Browser Not SupportedAn Online sEMG Motion Classification Framework for Tele-operating the Robotic Hand... doi.org/10.23919/CCC50068.2020.9188795Browser Not SupportedAn Online sEMG Motion Classification Framework for Tele operating the Robotic Hand doi 10 23919 CCC50068 2020 9188795
Read online
File size648.74 KB
Pages16
DMCAReport

Related /

ads-block-test