Trace:

Differences

This shows you the differences between two versions of the page.

--- projects:hand_cartesian [2010/08/24 17:32] – memeruiz
+++ projects:hand_cartesian [2010/08/24 17:36] (current) – memeruiz
@@ Line 1: / Line 1: @@
+====== DLR/HIT hand cartesian library ======
+The hand cartesian library enables the DLR/HIT hand to:
+  * Calculate current cartesian position of finger tips.
+  * Calculate current forces at finger tips.
+  * Run different types of controllers for each finger.
+  * Close loop inverse kinematics controller.
+  * Velocity inverse kinematics.
+===== Installation instructions =====
+  * Get the code from oid5 repository
+  * Compile patched version of orocos-kdl
+  * Compile yarp
+===== Prerequisite: Run the DLR/HIT server =====
+  * Run the server for the DLR/HIT hand in the computer where the hand is connected:
+  cd oid5/robots/kimp/hand/sahand_api
+  ./sahand_yarp.py -n
+===== Usage =====
+  * Creating the hand object:
+  import hand as h
+  hand=h.Hand(handedness="left",sahand_number=1,portprefix="",sahand_port_name_prefix="/sahand")
+Use the values adequate for your configuration.
+sahand_number indicates the port where the hand was connected to the PCI card.
+sahand_port_name_prefix indicates the prefix used by the DLR/HIT server.
+handedness is used for selecting the correct kinematic description.
+portprefix is used to add a prefix to the client port names.
+  * Getting data from the hand:
+  hand.update_sensor_data()
+  first=hand.fingers[1]
+  print first.cur_joint_pos
+This prints the current joint position of the finger.
+The finger class has a lot of more members that contain data or functions to calculate.
+  * To get access to the kinematics members of the finger just look into fingerobject.kinematics.*
+  * To get access to the controller members of the finger just look into fingerobject.controller.*
+  * To control the hand. First select the controller for each finger that you want to use:
+  hand.set_controller(Finger_controller_sahand_joint_pos,finger_list=[1])
+First argument is a controller class in this case Finger_controller_sahand_joint_pos is the controller that only relays information to the hand directly. There are other controllers like a close loop inverse kinematics one.
+The second argument is a list of fingers which will have this controller.
+  * To command a finger to move to a joint position:
+  from numpy import pi
+  hand.set_refs(1,[0,10*pi/180,0])
+  hand.update_controller_refs()
+Angles are in radians. Any other units are the standard metric scale (meters, seconds,Nm, etc)
+To know what the refs are, please take a look in the set_refs member in the selected controller class.
+The joint order is: first in the list corresponds to the first joint of the kinematic chain starting from the base of the chain.
+  * One can also change the controller parameters in the same than the references using hand.set_params(finger,params_list) and then hand.update_controller_params()