Collaborative Workshops: AI Limb Gripping Point Prediction and Path Execution Achieved
23. februar 2022
Between the 31st of January and the 9th of February, there has been held a series of RoBUTCHER workshops at NMBU. This workshop series consisted of a mixture of remote support and physical attendance and included attendance at NMBU by the following consortium members:
ANIMALIA: Per Håkon Bjørnstad
OBUDAUNI: Kristóf Takács
BYTEMOTION: Haris Hadzic, Tommy Jonsson
RobotNorge: Håvard Bakke
CIKLUM: Anton Popov, Oleh Smolkin
NMBU: Alex Mason, Steven Ross, Luis Eduardo Cordova-Lopez, Ian de Medeiros Esper, Dmytro Romanov, Vebjørn Corneliussen Storvik
The list of testing activities shown below is a testament to the successful collaborative work carried out throughout this series. The highlight of the workshops has to be the success of the AI gripping point prediction and path execution that result in sound manipulation of the LH shoulder by the robot gripper in conjunction with cutting by a butcher for the limb removal.
Activities and achievements
Tested CHU updates.
Testing of new hook pluck holder for inspection rack.
Calibration of the systems and tools executed, with new calibration methods also under investigation by RobotNorge for future workshops.
BYTEMOTION and ROBOTNORGE have shown that the Node-Red system used to coordinate and control the whole process is making good progress.
CIKLUM’s AI has undergone more testing and improvements, and automatic gripping of the legs using AI predicted gripping points is now possible and demonstrated focussing mainly on LHS shoulder removal.
AI is also proven to work well for predicting the LHS ham gripping point.
OBUDAUNI’s new upgraded gripper was used throughout the workshop and tested for functionality and payload capability.
Trachea cutting using the manually driven cutting robot Dexter proves that it is possible with no reach problems observed.