Re Fibres - Using advanced deep learning, computer vision models, multispectral sensors & automated robotics to develop an end-to-end UK recycling framework for textile waste

• InnovateUK
• May 2024 - May 2026
• Total Budget £1.68M, KU: £121,000

Re Fibres - Using advanced deep learning, computer vision models, multispectral sensors & automated robotics to develop an end-to-end UK recycling framework for textile waste

First Hardware Integration and Digital Twin

• The first integrated platform with the robotic arms is available
• They are operational but in controlled experiments and a small variety of textile samples
• The Digital Twin of the robotic arms is integrated also in the Unity3D environment
• Robotic Arm Digital Twin Design and Training
• The plans for the training and testing are in progress and expected to completed soon

Robotic Arms installation and testing

Robotic Arm Textile manipulation and integration

Initial Setup

• Robotic arm models: 2 x DOBOT Magician
• Conveyor belt: 2 x DOBOT Conveyor Belt Kit
• Grippers and end-effectors are attached to the robotic arms, to grasp and manipulate textiles without damaging them
• Depth sensors (Kinect) are integrated in the setup to provide real-time feedback on the position and orientation of the textiles
• The plans for the training and testing are in progress and expected to completed soon

Robotic Arms installation and testing

Unity Digital Twin

• The digital twin was integrated in the Unity3D environment and initial tests with ML-Agents were performed.
• The DT supports all the degrees of freedom that the actual DOBOT MagicianIt supports
• Next tasks include the completion of the training process and provide a first evolution study with numerical results
• We aim to measure accuracy for the task completion and the robotic arm's adaptability