A CAD based supervised bin-picking real-time Object Detection system
The solution could be properly and efficiently used in a bin picking industrial scenario to identify and localize, using a 2D approach, the best object to pick among a multitude of identical ones.
The solution is suitable for that cases in which there is an a priori information of the objects to recognize and localize in the bin. The algorithm works in Real-Time providing a Fast Object Recognition inference time.
- State-of-the-art and real-time perception system for bin picking.
- ROS bindings.
- Detects 2D objects from a 2D image of a bin.
- High precision and execution speed.
- Agnostic to any sensor that produces 2D data.
- Need an apriori information of the objects (supervised learning).