Importing Generative AI into Human-Robot Collaboration, Promoting Intelligent Changes in Industries.

• Development of An Adaptive Multimodal Collaborative Surgical Robot System

This surgical robot is primarily used for percutaneous puncture surgeries, such as biopsy and tumor ablations. The system assists physicians in preoperatively planning puncture paths and precisely locates the puncture needle during surgery. Considering the deformable nature of soft tissues, it employs CT and ultrasound fusion imaging to monitor the puncture process in real time, thereby enhancing puncture safety and accuracy.

1. Generative AI, combined with physicians' clinical experience, establishes a needle insertion decision-making model.
2. CT and ultrasound image fusion technology provide real-time information during surgery, preventing soft tissue deformation that may lead to puncture errors and surgical risks.
3. Precise control by the robotic system reduces the difficulty of surgery for physicians and decreases the number of CT scans for patients.

• Human skill transfer for intelligent deburring with dual-arm robot

This project aims to develop an automated deburring system using dual-arm robots and force modeling to transfer craftsmen's skills. Combining advanced robotics with human expertise, we seek to enhance deburring quality and efficiency in high-precision manufacturing.

The advanced force control technology developed in this project combines expert craftsmanship to outperform traditional methods in machining precision. It includes burr detection for machining inspection and quality management, enhancing efficiency and quality while supporting Industry 4.0 standards, boosting Taiwan's manufacturing competitiveness.

• 5G-AI Human-Machine Collaboration Technology Using High-Precision Full-Field Stereo Vision and Heterogeneous Sensor Fusion for Personnel Posture Prediction and Robotic Arm Path Planning

The various technologies developed in this integrated project possess versatility and broad applicability. For different human-machine collaboration workflows and different product manufacturing/assembly processes in various fields, it is able to make specific parameter adjustments tailored to the enterprise.

1. Laser Galvo System - Assisting Operator Target Recognition
2. Human 3D Skeleton Recognition - Operator Action Analysis and Data Collection
3. Dynamic Real-Time Full Arm Obstacle Avoidance - Arm Workspace Safety Detection
4. Large Language Model - Achieving Seamless Human-Machine Interaction

• Integration of human-machine collaboration industrial robots and autonomous vehicles in casting-part polishing

1. Automated Polishing Quality Inspection and Workpiece Positioning: Utilizing deep learning image technology to assess whether a casting can be polished. The casting is taken to the processing platform for positioning, and the actual polishing trajectory is adjusted according to the positioning results.
2. Active Contact Flange Polishing Force Control: During the polishing process, as the surface material is removed, the arm needs to adjust the trajectory according to the polishing normal force to achieve consistent force, so that the surface polishing has the best uniformity.
3. Generative AI for Robotic Polishing Systems: Utilizing texts to give commands to the robot arm, and the robot can make semantic inferences to complete the action task.
4. Castings, water hardware, and furniture polishing industries: Utilizing the electromagnetic active contact device developed in this project can provide force control accuracy below 1N.
5. Casting and PCB inspection industry: The dual robotic arm automated optical inspection image recognition technology introduced in this project can provide different camera inspection angles, and the accuracy is expected to reach more than 90%.
6. 2D vision workpiece positioning technology: The introduction of this project technology can use a low-cost 2D camera to estimate the 6D posture of the workpiece.

• Delopment of Collaborative Robots with High Dynamic Impedance Response for Intelligent Human-Robot Cooperation

A new collaborative robot with high dynamic compliance response is proposed. Without sacrificing the rigidity, repeatability, payload-to-weight ratio, and cost, the new robot can satisfy the rapid active and passive compliance motion required by human-robot cooperation.