Bachelor's Thesis. Is ROS for me? A non hardcore programmer
Hi!
I have a Bachelor's thesis with an end-effector on a robot, KUKA KR210. I am mostly an mechinical engineer but i start to get intrested in programming and robots. For this project i am thinking about having cameras on the end-effector as a safety feature to give the end-effector "eyes" and can act as an emergancy stop if a human is to close. This end-effector is used in an industry production line and needs to be safe for a fitter to work along side with
Is this possible to implement? If yes, how should i do it?
Note; i am very new to this am afraid, but really like to learn and make it work
I don't want to discourage you, and for a BSc it's probably ok, but a camera connected to a RPi with ROS is never going to get safety-rated. What you are describing ("emergency stop if a human is to close [..] needs to be safe for a fitter to work along side with") is definitely possible ..
.. but please consider that companies in this line of work (ie: safety-lasers, co-bots) spend tens of thousands of dollars developing and getting their equipment certified. A KR210 is easily capable of killing a human, so please be careful: never make your RPi the only system.
Thanks for the reply!
I see, so to get safety-rated level there's a need for a more powerful computer? Or is it the concept with cameras and the pictureprocessing that makes the reactiontimes to slow to be used as an emergency stop?
No, my point was that making software and hardware really suitable for safety critical applications (which acting as an emergency stop is an example of) is hard and difficult. I'd recommend you try and talk to someone with knowledge in this field before you start.
But this all depends on ..
.. what the scope of your work is: if this is a one-off experiment for a BSc, and you're tutored by knowledgable people and your KR210 has additional safety measures in place, then it could be ok to test with your ROS application.
But don't put this in a production system -- with uninformed ..
.. users -- where there is a risk of really hurting people or damaging equipment.
Again, scope is important here, but I just wanted to point to some possible dangers.
and finally:
there are some experimental computer vision based systems that do something similar to this, but 1) I don't know whether they've already been certified and 2) "getting safety-rated" is probably not doable for you.
The scope, a robot picks up a plasticpanel and then place it under a car, then a fitter will assemble it with screws. The car will move with a constant speed and the robot with the panel will move along. So we need to make the layout and the end-effector safe for the fitter with ..