Roles Founding Pro, Tactile Sensor Beta Testers
Full Name Matthew Bush
Job Title COO/Cofounder
Country US - United States
Company (optional) Hirebotics, LLC
I'd agree with picking and packing directly. I'd look to see if there is a way to spread out the product on the bed. Do the sachets come out at different points onto the belt? If so you may want to have those points discharge at different points on the width of the belt so that the product is more spread out.
I think you're going to need vision to be able to orient the packages correctly. What type of robot are you looking to use at that speed? You'll probably need either a delta or scara and then probably want a couple of them to ensure everything is picked.
We use a standard naming convention of CustomerName_ProjectName_Default for the default program that loads automatically. We then store our versioned programs in an archive folder. In that folder we use semantic versioning with major.minor.bugFix. We use the naming convention of CustomerName_ProjectName-vx.y.z
The CustomerName and ProjectName we abbreviate so that it's not too long, generally 3-5 letters for the customer and project names.
In the before start section of the program we use a variable named VERSION to store the current version number in so that appears near the top of the variables list. Helps us to identify quickly with customers if they are running the correct version of the program.
We do USB magic file backups at the end of each day or programming and sometimes more frequently depending on what's going on. We also have the ability to backup programs to AWS remotely so we can grab files at anytime to look at.
The short answer is yes, you should be able to do that. You would need to set-up a plane reference and then set your pallet to that plane feature so that adjusting the plane would adjust the palletizing sequence. In your program the robot would go to each of the plane points and search until you found the point, you would then reset the plane point to be the actual TCP location you just found. The 3 points that make up the plane would need to be variable points so that you could programmatically change the points. Once the points are relearned the plane feature would be adjusted automatically and should then adjust the palletizing sequence to the new points.
We have a couple of areas where I think time and money are wasted but today are necessary evils.
Part presentation - the ability to handle parts presented in bins the way humans would pick them would reduce lots of system design time as well as cost. Part presentation systems can run from a few hundred dollars for something basic to tens of thousands for custom feeders and can add months to the project implementation time waiting for them to be designed and manufactured.
Sensors/wiring - today most sensors are wired which means we have to run all the wires to give the cell it's nervous system. The cell I'm deploying this week has roughly 20 sensors in it in various parts so we can give the robots all the data they need to know how to run the cell efficiently.
Gripper Finger Design - because robots today lack tactile feel we have to design a lot of custom fingers for grippers whether we are using a pneumatic or robotic gripper. It's the rare case that we feel confident deploying with off the shelf finger tips as we need to ensure the part is picked correctly and accurately so use mating features to make sure we know where the part is relative to the gripper.