Abstract:This paper presents a data-driven kinematic model for the end-effector pose control applied to a variety of manipulator robots, focusing on the entire end-effector's pose (position and orientation). The measured signals of the full pose and their computed derivatives, along with a linear combination of an estimated Jacobian matrix and a vector of joint velocities, generate a model estimation error. The Jacobian matrix is estimated using the Pseudo Jacobian Matrix (PJM) algorithm, which requires tuning only the step and weight parameters that scale the convergence of the model estimation error. The proposed control law is derived in two stages: the first one is part of an objective function minimization, and the second one is a constraint in a quasi-Lagrangian function. The control design parameters guarantee the control error convergence in a closed-loop configuration with adaptive behavior in terms of the dynamics of the estimated Jacobian matrix. The novelty of the approach lies in its ability to achieve superior tracking performance across different manipulator robots, validated through simulations. Quantitative results show that, compared to a classical inverse-kinematics approach, the proposed method achieves rapid convergence of performance indices (e.g., Root Mean Square Error (RMSE) reduced to near-zero in two cycles vs. a steady-state RMSE of 20 in the classical approach). Additionally, the proposed method minimizes joint drift, maintaining an RMSE of approximately 0.3 compared to 1.5 under the classical scheme. The control was validated by means of simulations featuring an UR5e manipulator with six Degrees of Freedom (DOF), a KUKA Youbot with eight DOF, and a KUKA Youbot Dual with thirteen DOF. The stability analysis of the closed-loop controller is demonstrated by means of the Lyapunov stability conditions.

Enhanced Kinematic Model for Dexterous Manipulation with an Underactuated Hand

Dexterous Manoeuvre Through Touch in a Cluttered Scene

Contact-Implicit Model Predictive Control for Dexterous In-hand Manipulation: A Long-Horizon and Robust Approach

Inverse Kinematics Solving of a Novel Soft Manipulator with Vision and IMU Data

Robust Dexterous Manipulation Under Object Dynamics Uncertainties

Non-Parametric Self-Identification and Model Predictive Control of Dexterous In-Hand Manipulation

Dexterous Manipulation Control of a Bionic Prosthesis in Cooperation with Human Upper Limbs

Deep Dynamics Models for Learning Dexterous Manipulation

A High-Efficient Reinforcement Learning Approach for Dexterous Manipulation

Hybrid Learning- and Model-Based Planning and Control of In-Hand Manipulation

Complementarity-Free Multi-Contact Modeling and Optimization for Dexterous Manipulation

Object-Centric Dexterous Manipulation from Human Motion Data

DexSim2Real$^{2}$: Building Explicit World Model for Precise Articulated Object Dexterous Manipulation

Humanoid Grasping with Multi-Finger Dexterous Hands Based on DM-KMP

The Role of Learning and Kinematic Features in Dexterous Manipulation: A Comparative Study with Two Robotic Hands

Dynamic Modeling and Analysis for Robotic Arm-Hand Systems

Enhanced Robust Motion Control based on Unknown System Dynamics Estimator for Robot Manipulators

Sampling-Based Model Predictive Control for Dexterous Manipulation on a Biomimetic Tendon-Driven Hand

Robust In-Hand Manipulation with Extrinsic Contacts

Data-Driven Kinematic Model for the End-Effector Pose Control of a Manipulator Robot

An Under-Actuated Manipulation Controller Based on Workspace Analysis and Gaussian Processes.