I have a Robotic arm mounted on a car. There's a camera attached to it. Suppose the camera takes the image of a room, and finds that there's something, say an object, that has to be picked up. Say it's 50 feet away from the robot. My question is that how will the robot reach the object in the first place, and secondly, when it has reached the object, how will it know the real world co-ordinates of the object, to pick the object up, using inverse kinematic equations. Any help would be appreciated. Thanks

  • $\begingroup$ Welcome to Robotics.SE! Your question is very very broad. All the things you mention by the way: "the robot senses that there is something", "makes the decision that the 'something' should be picked up", "calculates the distance of 50 feet", "moving the robot", "moving the robot to the found position", "moving a robotic arm with IK" each of them fills entire rooms of libraries with books on how to do it, because they are huge fields of research on their own. Please be more specific. $\endgroup$ Jul 8 '15 at 14:48
  • $\begingroup$ Yes, please break this into at least 2 questions. (mobile robot navigation, and object pose detection). Do a little research on your own first, then come back when you get stuck and have more specific questions. $\endgroup$
    – Ben
    Jul 8 '15 at 17:23
  • $\begingroup$ "the real world co-ordinates" which world?? $\endgroup$
    – CroCo
    Jul 10 '15 at 7:44
  • $\begingroup$ I've tried to ask only a part of the question here robotics.stackexchange.com/questions/7682/…. Any help would be appreciated $\endgroup$ Jul 13 '15 at 1:02

With your question, i assume that you are new to this field. This cannot be solved with one concept.

The mentioned objective requires SLAM technique - Simultaneous localization and mapping .

Using SLAM, the robot can be navigated to any location. A camera is not enough to map an area or a room in your case. You also require depth sensors such as kinect or Zed. You need to develop a program to reach the object for your robot, depending on the architecture that you are using. Here comes the role of a platform. A typical arduino, or a raspberry pi single board computer is enough for your project. You should be very familiar with these platforms to integrate your sensors and play with them.

Finally, once the location is reached and the object is recognized by the sensors, Here comes the Inverse kinematics into the action. There is no such thing called "real-world co-ordinates", until and unless you are dealing with GPS for long range SLAM.

In your case, the co-ordinates of the object are measured in the context of the location of the sensor mounted on your car.

Once the parameters of the object are known, you can use IK(inverse kinematics) for your robotic arm to grab it from its position. (Assuming that you know about IK, as you have mentioned in the question)

  • $\begingroup$ "You need to develop a program to reach the object for your robot" isn't very helpful. $\endgroup$ Jul 8 '15 at 17:58
  • $\begingroup$ i meant, it is one of the things to do after mounting sensors depending on his architecture, I cannot be more specific for this broad question. $\endgroup$ Jul 9 '15 at 2:00
  • $\begingroup$ @Trinadhvenna, I don't believe the OP needs SLAM. The problem is not to localize the robot in an unknown environment. The problem is to search and detect an object and moves toward it. $\endgroup$
    – CroCo
    Jul 10 '15 at 7:42

Not the answer you're looking for? Browse other questions tagged or ask your own question.