Description of datasets: Reaching --------- The dataset is a number of demonstration of a robot performing a reaching motions (in R^3). The data recorded is the position in space of the end effector (the hand). Dimensions correspond to: 1 2 3 4 5 6 X Y Z Xdot Ydot Zdot The goal for this dataset is to learn the velocity profile over one of the dimensions (e.g. Xdot) as a function of the 3D position. This allows the robot to know "where to go" when it is in a specific position. Grasping --------- The dataset is a number of demonstration of a robot grasping an object. The input data is the response of pressure sensors on the robot's finger. Dimensions correspond to: Grasping Raw : 1-7 : arm joints velocity 8-13 : fingers joints velocity 14-25 : 12 raw index pressure sensors 26-37 : 12 raw middle finger pressure sensors 38-49 : 12 raw thumb pressure sensors Grasping Avg: 1-7 : arm joints velocity 8-13 : fingers joints velocity 14-16 : average index, middle finger and thumb pressures The goal for this dataset is to learn the velocity profile of the joints from the pressure sensors of the hand. This allows the robot to learn the best way to grasp an object. Group List ---------- Group Dataset Input Dims. Regression Dim. 1 Reaching1 1 2 3 4 5 2 Reaching1 1 2 3 5 6 3 Reaching2 1 2 3 4 5 4 Reaching2 1 2 3 5 6 5 Reaching3 1 2 3 4 5 6 Reaching3 1 2 3 5 6 7 Reaching4 1 2 3 4 5 8 Reaching4 1 2 3 5 6 9 Reaching5 1 2 3 4 5 10 Reaching5 1 2 3 5 6 11 Reaching6 1 2 3 4 5 12 Reaching6 1 2 3 5 6 13 GraspingRaw 14-25 (use 3) 1-4 (try 2) 14 GraspingRaw 26-37 (use 3) 1-4 (try 2) 15 GraspingRaw 38-49 (use 3) 1-4 (try 2) 16 GraspingRaw 14-25 (use 3) 5-9 (try 2) 17 GraspingRaw 26-37 (use 3) 5-9 (try 2) 18 GraspingRaw 38-49 (use 3) 5-9 (try 2) 19 GraspingRaw 14-25 (use 3) 10-13 (try 2) 20 GraspingRaw 24-37 (use 3) 10-13 (try 2) 21 GraspingRaw 38-49 (use 3) 10-13 (try 2) 22 GraspingAvg 14-16 1-4 (try 2) 23 GraspingAvg 14-16 5-9 (try 2) 24 GraspingAvg 14-16 10-13 (try 2)