Minimum jerk for human catching movements in 3D

Abstract

Abstract-To investigate fast human reaching movements in 3D, we asked 11 right-handed persons to catch a tennis ball while we tracked the movements of their arms. To ensure consistent trajectories of the ball, we used a catapult to throw the ball from three different positions. Tangential velocity profiles of the hand were in general bell-shaped and hand movements in 3D coincided with well known results for 2D point-to-point movements such as minimum jerk theory or the 2/3rd power law. Furthermore, two phases, consisting of fast reaching and slower fine movements at the end of hand placement could clearly be seen. The aim of this study was to find a way to generate human-like (catching) trajectories for a humanoid robot.

Incremental bootstrapping of parameterized motor skills

“…The reward depends on the DMP parameters θ that result in a 10 dimensional joint trajectory transformed by the kinematics of the robot arm to the end-effector trajectory p t . The jerk is based on the third derivative of the end-effector trajectory p t as proposed in [4] and is represented as one objective in the reward function Eq. 4(a).…”

The M-basis Functions and Their Application

“…Subject to knowing the boundary conditions, i.e., the values of θ and up to the (n-1) th derivative of θ at t = 0 and t = tf, we will show that the solution to (1) is a linear combination of 2n basis functions that we call M-basis functions. First, the third-order (n = 3) is discussed because (1) results in minimum jerk patterns [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16].…”

Motion Planning by Demonstration With Human-Likeness Evaluation for Dual-Arm Robots

“…hand jerk (i.e. the third time-derivative of the hand position) [16], joint jerk [17], joint torque [18] or a convex combination of several cost functions with weighting factors chosen to describe an observed human motion [19]. Hence, minimizing these cost functions, human-like motions can be obtained.…”