Robot Pose Estimation and Normal Trajectory Generation on Curved Surface Using an Enhanced Non-Contact Approach.

Area of Science:

• Robotics
• Manufacturing Engineering
• Computer Vision

Background:

• Robotic machining offers efficiency but faces challenges with curved surfaces, including fixture errors and surface friction.
• Existing non-contact and contact-based methods have limitations that hinder precise surface finishing.
• Accurate path following and normal trajectory generation are critical for high-quality robotic machining of complex geometries.

Purpose of the Study:

• To develop an advanced technique for path correction and normal trajectory generation in robotic machining of curved surfaces.
• To overcome limitations of prior methods, specifically fixture errors and surface friction.
• To enhance the precision and reliability of robotic surface finishing operations.

Main Methods:

• A key-point selection approach with a depth measuring tool estimates workpiece coordinates, mitigating fixture errors.
• An RGB-D camera on the robot's end-effector determines depth and angle, addressing surface friction.
• A pose correction algorithm utilizes point cloud data for guaranteed perpendicularity and constant surface contact.

Main Results:

• The proposed technique demonstrated superior normal trajectory generation compared to existing methods.
• Experimental trials using a 6-DOF robot manipulator validated the technique's effectiveness.
• Achieved average angle and depth errors of 1.8 degrees and 0.4 mm, respectively.

Conclusions:

• The advanced technique successfully corrects robot paths and generates accurate normal trajectories for curved surfaces.
• The integration of depth sensing and point cloud analysis effectively minimizes errors and friction.
• This method significantly advances the capabilities of robotic machining for complex surface finishing applications.