GE's team of field engineers provide manual maintenance services for industrial assets that GE manufactures and sells. This method results in crew safety concerns, longer assets downtime, lost revenue, and added operational expenses.
That's where you come in!
GE wants to build service robots to help its field engineers with asset inspection and repair. The primary objective here is to detect and manipulate an object of interest in industrial settings.
For robots to manipulate objects in an unstructured industrial environment, it needs to possess a way to find the object of interest, compute its grasp location gives the object pose and the end – effector shape and size, plan a trajectory or visually servo to reach the grasp location and then grasp the object. Three of the central challenges that need to be addressed for the successful execution of this task are: object tracking, pose estimation, and grasp location estimation.
In this 1st robotics challenge GE is looking for novel approaches for object detection and tracking within industrial settings. Traditionally computer vision based approaches are used for object detection and tracking, but outside-the-box solutions that utilize information from multiple modalities- visual, shapes, and semantic information- are highly encouraged.
Download and review the detailed challenge description PDF attached below for an overview of the challenge rules and scoring criteria. The PDF also includes information on obtaining and parsing the input data.
Best of luck!
[PDF = Everything from original PDF except Background section]
|Winner receives 195K|
|Leaders will be displayed here as they are rated...|