3. Robot Program¶

The example program implements a complete robot vision workflow: calibrating the camera to the robot, detecting objects, and moving to them. It is split into three URScript modules imported together with Generic wenglor interface.urpx.

Modules¶

Program flow¶

Each example routine first calls wenglor_helpers.calibrate_if_needed(), which runs a calibration if no calibration data is available on the device yet, then loads the relevant uniVision job and moves to g_detection_pose.

Calibration¶

The calibration process differs depending on whether the camera is mounted on the robot or not. The sections below describe only how the Polyscope X example performs each case.

The poses are taught as WG_CALIB_POSE_1 … WG_CALIB_POSE_5; wenglor_api.run_calibration moves through each of them in sequence, calling wenglor_api.add_calibration_pose (which issues calibration:add[...]) at every pose:

optimovej(... WG_CALIB_POSE_1 ...)
wenglor_api_add_calibration_pose()
optimovej(... WG_CALIB_POSE_2 ...)
wenglor_api_add_calibration_pose()
... (through WG_CALIB_POSE_5) ...
wig_command = "calibration:calculate[" + WG_USE_CASE + "," + WG_CALIBRATION_TARGET + "];"

Add further calibration movements in wenglor_api.run_calibration if you need more than five poses (see Troubleshooting).

Camera on robot¶

For WG_USE_CASE == "camera_on_robot", run_calibration sets the detection pose to the first calibration pose automatically once calibration succeeds:

if (WG_USE_CASE == "camera_on_robot"):
  g_detection_pose = WG_CALIB_POSE_1

Camera not on robot¶

For WG_USE_CASE == "camera_not_on_robot", run_calibration performs an additional ground-calibration step after the hand-eye calibration: it moves to g_detection_pose, waits for operator confirmation that the calibration target has been placed on the object plane, then sends calibration:ground[WG_CALIBRATION_TARGET].

Verification¶

wenglor_api.validate_calibration performs an optional verification step once a calibration is present on the device (the second bit of state[WG_USE_CASE] is 1):

1. Checks the camera/calibration state via wenglor_api.update_camera_status (which sends state[WG_USE_CASE]).
2. Prompts the operator and moves to g_detection_pose.
3. Sends validate[WG_USE_CASE, current_tcp_pose] to the vision device.
4. Moves the robot to the returned calibration-target pose, offset upward by WG_VALIDATION_OFFSET_Z_MM along the tool Z axis, so the operator can visually confirm the result.

Prerequisite: the calibration plate must be visible from g_detection_pose. Add a call to wenglor_api.validate_calibration in the Main Program to use it — it is not called automatically by the example routines.

Detection¶

After successful calibration, the program detects objects and moves the robot to the position the camera reports.

single_detection¶

Loads WG_FIND_OBJECTS_JOB, moves to g_detection_pose, sends detect[...], reads the shape model and additional value of the detected object, then moves the robot to the returned object pose:

1. Loads WG_FIND_OBJECTS_JOB (via wenglor_api.load_find_object_job).
2. Moves to g_detection_pose.
3. Sends detect[...] (via wenglor_api.detect_objects).
4. Reads shape model and additional value of the detected object (wenglor_api.read_shape_by_index, wg_shape_model, wg_additional_value).
5. Moves the robot to the returned object pose.

multi_detection¶

Same as single_detection, but iterates through every object returned by num_objects:get (via wenglor_api.read_num_objects), reading pose (wenglor_api.read_pose_by_index), shape, and additional value for each:

wg_object_index = 0
wenglor_api_detect_objects()
wenglor_api_read_num_objects()
while (wg_object_index < wg_num_objects_found):
  wenglor_api_read_pose_by_index()
  wenglor_api_read_shape_by_index()
  optimovel(wg_object_pose, ...)
  wg_object_index = wg_object_index + 1
end

update_reference_frame¶

See 4.6 Target Pose and Camera-to-Target Calibration in the wenglor robot vision manual for the underlying target:pose command.

1. Loads WG_FIND_TARGET_JOB, moves to g_detection_pose, and triggers a calibration-target detection via wenglor_api.detect_target (which sends target:pose[WG_USE_CASE, WG_CALIBRATION_TARGET, wig_tcp_pose_string]).
2. Creates (or updates) the frame w_ref_frame, attached to world.
3. On the first run (WG_MACHINE_POSES_TAUGHT == False), the program stops so the operator can teach G_POSE_IN_MACHINE relative to w_ref_frame. Set WG_MACHINE_POSES_TAUGHT to True and restart the program.
4. On subsequent runs, the robot moves to G_POSE_IN_MACHINE.

This workflow lets you re-localize a machine or fixture automatically between runs without re-teaching downstream poses.

Selecting a routine¶

Navigate to the program tab and call either single_detection or multi_detection from the wenglor_examples module as the Main Program entry point. Select the TCP used for calibration, execute the program, and monitor the messages displayed on the teach pendant.

Units and conventions¶

• Positions in poses (p[x, y, z, rx, ry, rz]) are in meters; rotations use the rotation-vector (Rodrigues) convention in radians — the same units used by the generic robot vision API, so wenglor_helpers.update_tcp_pose_string and wenglor_helpers.string_to_pose only need to format/parse the string, not convert units.
• wenglor_helpers.string_to_pose parses the comma-separated x,y,z,rx,ry,rz reply into a p[...] pose by locating each comma with str_find and slicing the string with str_sub.