The robot maintains its true pose in the variable truePose. Its current target pose is currentPose. It always drives its true pose to its target pose, and generally converges to within a millimeter.

You can push the current pose on the stack, then move the arm, and then assume the pose with moveEeToPoseWord. For example, the following program pushes the current pose, moves the arm to crane pose, then moves back to the current pose:

currentPose assumeCrane1 waitUntilAtCurrentPosition moveEeToPoseWord

When Ein is in zero gravity mode, the current position is constantly set to the true position, so the arm stays where you move it. Otherwise, if the arm is moved, it will move back to the current pose. You can toggle zero gravity with zeroGOff and zeroGOn or pressing the circular grey button on the arm (the “OK” button).

Additionally, Ein can be configured to stop publishing movement commands. Call publishCommandsOff to tell Ein not to publish commands to the robot, so that e.g., MoveIt can move the robot. You can use publishCommandsOn to move the robot again.

publishCommandsOff and zeroGravityOn are similar. However zeroGravityOn means that Ein will continue to publish position messages; they will just be set to the true pose. publishCommandsOff means Ein will not publish any messages at all.

Exercise: Play with Zero Gravity.

Toggle zero gravity mode and move the robot’s end effector by squeezing the wrist cuff. Try to pick up objects using the buttons on the arm to open and close the gripper, and explore the workspace of the robot. Turn off zero gravity, and observe what happens when you move the end effector.

Answer (select to see):

When zero gravity is on, Ein will stay when you move the end effector, because it sents the current position to the last observed position. When it is off, it will move the end effector to the last position it was at.

Exercise: Play with publishing commands.

Turn off publishing commands and then use the joint angle control to move the robot. Use: rosrun baxter_examples joint_position_keyboard.py from a bash workspace. What happens if you run it with publishing commands turned on? What if it is off?

Answer (select to see):

When publishCommandsMode is on, the rethink program and Ein will “fight” and Ein will return the end effector to the true pose. When it is off, you can use the joint position program to move the end effector and it will stay there.

Stopping

waitUntilAtCurrentPosition causes the robot to pause until the end effector is within a threshold of its target position. comeToStop will pause until the robot has actually reached a stop (waiting until the current position has stabilized to within a threshold). The properties w1GoThresh and w1AngleThresh control how close `waitUntilAtCurrentPosition must get before deciding it is at the current position; they have equivalent setters.

Exercise: Waiting

Change w1GoThresh and observe how close waitUntilAtCurrentPosition needs to get while stopping. While changing this value, set w1AngleThresh to a very large number to avoid conflating position and angle. For very small values for w1GoThresh, waitUntilAtCurrentPosition may never return.

Default value: 0.003 setW1GoThresh zUp waitUntilAtCurrentPosition

Takes noticably longer: 0.0005 setW1GoThresh zUp waitUntilAtCurrentPosition

Never terminates: 0.000005 setW1GoThresh zUp waitUntilAtCurrentPosition

Grid Size

The grid size determines how much the arm moves for the various commands like xUp. It is set in centimeters, so you can use 0.01 setGridSize so that each xUp moves 1 centimeter, or 0.1 setGridSize so that each xUp moves ten centimeters.

Movement Speed

Try moving a long distance, like 20 cm. How fast does the robot move? To move slowly, issue

0.05 setSpeed

The speed is specified with a number between 0.0 and 1.0. There are some shortcut words for certain speeds:

fullImpulse
halfImpulse
quarterImpulse
tenthImpulse

Large movements should generally be carried out in quarterImpulse or tenthImpulse. Speeds greater than or equal to halfImpulse or 0.5 should be used with caution; even the safest robot demands attention.

Exercise: Try different speeds.

Try moving the arm up and down with different movement speeds and grid sizes.

Changing the Home Position

You can change the home position with build in getter and setter commands. Remember to set separate values for each arm. First drive the arm to the desired new position. We recommend using a point in the green part of the IK workspace so the arm can move freely around the home position for servoing and mapping and the like. Issue

beeHome print
currentPose setBeeHome
beeHome print
currentPose print

and then move the arm somewhere else. Run

currentPose print
goHome
currentPose print

and verify the output is what you would expect. If you would like the home position to be saved across different Ein restarts, you can add these commands to init.back. They will then be executed at startup.

Force feedback

Ein can use the robot’s torque sensors to press down until it senses force on the table. To do this feedback loop, move the gripper until it is at a position as close as you can manage to the object. Moving down is slow, because the robot must sense torque at each step. The robot will gradually move down until it senses a force. To call the “all in one” word, use pressUntilEffortAllInOne which first comes to stop, then initializes variables, then slowly moves until it touches. It will then back up slightly to unwind the springs and then gently move down to touch the table. It uses localZDown to move, so make sure it is in crane pose if you wish to touch the table.

The word pressUntilEffortCombo touches the table coarsly, then backs up and presses again until it is lightly touching the table. It must be called after calling pressUntilEffortInit which sets various variables needed for pressUntilEffort. You can call pressUntilEffotInit to set up default values, and then individually set specific things, such as the movement speed.

The following parameters are set in pressUntilEffortInit:

  • setSpeed which sets the arm’s movement speed.
  • setGridSize which sets the grid size during movement.
  • setW1GoThresh which is the distance that waitUntilCurrentPosition waits to be within before deciding it is at the current position.
  • setEffortThresh which is how much force will trigger a detection of effort.

Exercise: Press on different surfaces.

Use pressUntilEffortAllInOne to press on different surfaces, and observe how much force is needed for various actions. Change effortThresh and observe how that affects the force.

Exercise: Explore the press until effort parameters

Use pressUntilEffort and pressUntilEffortInit to explore the various parameters. Try different speeds, grid sizes, and W1 thresholds and effort thresholds.