Any one can advice me on the ideal perception sensors for pick and place application using a robotic manipulator with ROS support. I have generally looked at things like Kinect and stereo cameras (Bumblebee2) which provide depth that can be used with PCL for object recognition and gripper positioning. Are there any other sensors would be preferred for such application and if not what are the drawbacks of stereo cameras in comparison to Kinect or other sensing capability.
The Kinect is certainly a popular choice these days for robotics. However, time-of-flight, structured light, and stereo cameras all have their own strengths and weaknesses. These two threads have a good discussion:
I think the simple answer is that stereo requires visual texture in the scene, and you can alter your baseline for different min/max ranges. TOF cameras might have trouble with certain materials, and can give you false readings when an object is too close to the camera.
The Structure sensor is a "3D sensor for mobile devices". Reading over the details it sounds like a smaller version of Kinect. It has open source drivers, openNI support etc. It's been on my to buy list for awhile but I don't have an immediate application just yet.
From the site,
The magic of 3D depth sensing begins with the ability to capture fast, accurate, dimensions of objects and environments.
I'm am in no way associated with the company that sells the sensor.
Kinect is an active 3D depth estimation setup that employs IR laser
structured patterns for Depth calculation and are reliable for Indoor
Stereo Camera setup is a passive 3d depth estimation setup that
typically involves use of Stereo calibration procedure in order to
compute projection matrix that will transform a 2D point(generally
observed in left camera) into a 3D point in the left camera coordinate
system(assuming left camera is at origin[0,0,0]). These stereo camera
setups performs way better than Kinect in case of outdoor