Vision guided robotics

Our research scenario

• Our interests lie in handling non-rigid objects such as fruit, with variable size, shape, firmness and texture.
• We have designed a robot gripper tool for grasping a fruit from its sides.
• A depth image created from a stereo pair of images from a calibrated camera is processed by a slice analysis algorithm, to identify individual objects.
• The pose of each object is obtained from a blob analysis at a specified depth.
• Candidate pick points are identified, from which the best is selected and its coordinates are calculated and sent to the robot.
• The robot picks up the fruit and transfers it to the delivery point.

A robot fitted with a multi-fingered gripper tool is used to grasp fruit from a pile

Robotic bin picking

• The automatic grasping of parts in a bin with the use of a robotic arm is commonly known as robotic bin picking. It is a complex task that is usually carried out by a vision guided robot (VGR).
• Recent developments in VGR have been in the vision technology. Now with low-cost 3D vision, complex issues can be overcome with novel approaches in image processing and analysis.

In this image, the positions of fruit have been identified and their pose has been determined. Suitable pick-up points are shown in black. The final pick point is in red