The Vision Interface

The Vision Interface has some variables and functions to communicate between Vision and robot. This interface should be able to communicate with different type of robots, no matter what language the robot is scripting in.

These are the examples:

Links below to API usage examples:

• Calibration Program
• Picking Program

Note

This article assumes an existing implementation of the socket communication layer.

Variables

Global Common Variables
daoai_socket_name	This is the name if socket which is “daoai” at all time.
daoai_tcp_pose	This variable is a vector which contains position and rotation p[0,0,0,0,0,0,0].

TCP pose vector contains the positions x, y, z and rotations(different robot uses different length of this vector, for example, UR uses 6 of them; ABB uses 7 of them).

Vision Variable
daoai_status	This variable is the command from Vision.

daoai_object_type & daoai_remaining_obj & daoai_payload: These are the 6 payloads of Vision replies. daoai_object_type is referred to payload_1 , daoai_remaining_obj is payload_2 , daoai_payload_x are referred payload -> 3, 4, 5, 6.

Robot is able to check this variable after receiving message from Vision . The status essential in communication since it is to determine which process Vision is currently in. It also represents the result from previous command. For example, if robot is under picking mode, robot requests the objects location, Vision has different reply based on its detection results: object found, object not found and image capture failure. Hence, this status is important to determine the current process between Vision and robot.

Robot Variable
payload_1 & payload_2	These are the payloads of the message.
daoai_r_command	This variable is the command from robot.

Robot sends command with/without poses to Vision , according to this command, Vision is able to determine what is the message and how to process this message. This is useful when robot is running on different script with Vsion flowchart: if robot is trying to perform calibration waypoints, but Vision is trying to detect object and perform picking; their commands are different in this communication. So that robot is able to detect the pose sent back from Vision is in wrong process. This is able to protect unexpected errors and collision.

For more details about these commands, please see Socket Communication.

Functions

All functions below are request to Vision, the communication is one send and one recv.

Warning

These are the system functions, do NOT directly call these functions even if you need to modify the script!

daoai_open_socket

Robot opens up the TCP socket via socket settings(port number and IP address).

send_robot_data

Robot sends the current pose and command to Vision.

recv_daoai_data

Robot receives status and corresponding poses from Vision.

daoai_close_socket

Robot closes DaoAI socket and terminates the running program.

These are the functions mainly for Manual Calibration process.

daoai_start_manual_calibration

Trigger the vision to start the flowchart of calibration.
Robot sends RC_START_MANUAL_CALIBRATION to Vision to request the start of Manual Calibration.

daoai_accumulate_calibration

Accumulate the calibration (will need to send the pose to Vision)
Until collected all the waypoints, robot calls daoai_stop_manual_calibration()
to terminate the process.

daoai_stop_manual_calibration

Robot sends RC_STOP_MANUAL_CALIBRATION to Vision to terminating calibration

These are the functions mainly for Guidance Calibration process.

daoai_accumulate_calibration

Accumulate the calibration, and once the vision received enough pose,
it will finish the calibration automatically

This is the same function as Manual Calibration

These are the functions mainly for Auto Calibration process.

daoai_start_auto_calibration

This is requesting Vision to start Auto Calibration. Robot sends RC_START_AUTO_CALIBRATION,
as well as the current robot pose. Then wait vision to send back acknowledgement.

daoai_is_calibration_done

Wait the response from the Vision, and write the pose into one object, if the return is no, exit

daoai_auto_accumulate_calibration

Accumulate the calibration (will need to send the pose to vision, and wait vision to send back
DAOAI_DONE_AUTO_CALIBRATION)

These are the functions mainly for Picking process.

daoai_find_objects(payload_1, payload_2)

Robot requests Vision to capture image and detect objects in the scene

daoai_receive_result

Robot waits for Vision side finishing the processing, and write the output to the pick pose

daoai_next_object(payload_1, payload_2)

Just receiving poses, without capturing new images. This can be applied to 1 capture and pick all
occurrences as well. payload_1 is the number of remaining objects in scene.