Dataset for: "Incremental Semiparametric Inverse Dynamics Learning"

Dataset used in the experimental section of the paper: R. Camoriano, S. Traversaro, L. Rosasco, G. Metta and F. Nori, "Incremental semiparametric inverse dynamics learning," 2016 IEEE International Conference on Robotics and Automation (ICRA), Stockholm, 2016, pp. 544-550. doi: 10.1109/ICRA.2016.7487177 Abstract: This paper presents a novel approach for incremental semiparametric inverse dynamics learning. In particular, we consider the mixture of two approaches: Parametric modeling based on rigid body dynamics equations and nonparametric modeling based on incremental kernel methods, with no prior information on the mechanical properties of the system. The result is an incremental semiparametric approach, leveraging the advantages of both the parametric and nonparametric models. We validate the proposed technique learning the dynamics of one arm of the iCub humanoid robot. URL: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7487177&isnumber=7487087 Description The file "iCubDyn_2.0.mat" contains data collected from the right arm of the iCub humanoid robot, considering as input the positions, velocities and accelerations of the 3 shoulder joints and of the elbow joint, and as outputs the 3 force and 3 torque components measured by the six-axis F/T sensor in-built in the upper arm. The dataset is collected at 10Hz at as the end-effector tracks circumferences with 10cm radius on the transverse (XY) and sagittal (XZ) planes (For more information on the iCub reference frames, see [4]) at approximately 0.6 m/s. The total number of points for each dataset is 10000, corresponding to approximately 17 minutes of continuous operation. Trajectories are generated by means of the Cartesian Controller presented in [5]. Input (X) columns 1-4: Joint (3 shoulder joints + 1 elbow joint) positions columns 5-8: Joint (3 shoulder joints + 1 elbow joint) velocities columns 9-12: Joint (3 shoulder joints + 1 elbow joint) accelerations Output (Y) Columns 1-3: Measured forces (N) along the X, Y, Z axes by the force-torque (F/T) sensor placed in the upper arm Columns 4-6: Measured torques (N*m) along the X, Y, Z axes by the force-torque (F/T) sensor placed in the upper arm Preprocessing - Velocities and accelerations are computed by an Adaptive Window Polynomial Fitting Estimator, implemented through a least-squares based algorithm on a adpative window (see [2], [3]). Velocity estimation max window size: 16. Acceleration estimation max window size: 25. - Positions, velocities and accelerations are recorded at 9Hz and oversampled to 20 Hz via cubic spline interpolation. - Forces and torques are directly recorded at 20Hz. This dataset was used in [1] for experimental purposes. See section IV therein for further details. For more information, please contact: Raffaello Camoriano - raffaello.camoriano@iit.it Silvio Traversaro - silvio.traversaro@iit.it References [1] Camoriano, Raffaello; Traversaro, Silvio; Rosasco, Lorenzo; Metta, Giorgio; Nori, Francesco, "Incremental Semiparametric Inverse Dynamics Learning", eprint arXiv:1601.04549, 01/2016 [2] F. Janabi-Sharifi ; Dept. of Mech. Eng., Ryerson Polytech. Univ., Toronto, Ont., Canada ; V. Hayward ; C. -S. J. Chen, "Discrete-time adaptive windowing for velocity estimation", IEEE Transactions on Control Systems Technology, 1003 - 1009, Vol. 8, Issue 6, Nov 2000 [3] https://github.com/robotology/icub-main/blob/master/src/libraries/ctrlLib/include/iCub/ctrl/adaptWinPolyEstimator.h [4] http://wiki.icub.org/wiki/ICubForwardKinematics [5] U. Pattacini; F. Nori; L. Natale; G. Metta; and G. Sandini; “An experimental evaluation of a novel minimum-jerk cartesian controller for humanoid robots,” in Intelligent Robots and Systems (IROS), 2010 IEEE/RSJ International Conference on, Oct 2010, pp. 1668–1674