Research – Vision-Guided Robotics – Summary
Vision Guided Bin Picking
Summary: (NSERC, Braintech, Precarn)
This is a large project with both industrial and research foci.
On the industrial side: this project will develop an intelligent system to tackle the problem of recognizing, locating, and picking automobile parts in a random pile in a storage bin. The outcome will be a solution to a long unsolved problem for robotic manufacturing where robots cannot yet distinguish parts randomly stored in a parts bin necessitating expensive structured industrial bin storage. While the initial project is directly sponsored by organizations in the automotive industry the intelligent system will have applicability in many other manufacturing sectors.
The research problems being considered include:
Methodologies for visual sensor (camera) placement/prepositioning as part of a fast, automated calibration of the system prior to initiating on-line operation. These methodologies would be applicable to a wide range of visually guided robotics tasks.
Development and implementation of robust strategies for target part searching and recognition. In particular, generation of motions that are collision free, avoid occlusions of proposed hypothesis targets and keep the targets within the camera field of view.
Fast and robust visual servoing techniques that provide smooth and efficient motion based on a minimum number of visual data updates. Smooth motion is essential to providing a minimally vibrating moving platform for image acquisition. Fast motion is important to meet industrial targets necessary for robotic bin picking to be useful.
Implementation and integration of robust feature tracking as part of the visual servoing scheme. This issue has not been as carefully considered in the context of visual servoing and is an important tool for ensuring successful bin picking.
Grasp selection, namely formulating an evaluation metric for determining the best part to pick. To achieve this, we are exploring how various factors affect “pick quality”
For supporting company information please see http://www.braintech.com/press_11_30_06.html
