Bulk sequencing (part picking) for automotive manufacturing

Community Projects
1

Hi everyone!

I am working on a project for an automotive manufacturing plant to try to improve their Bulk Sequencing (aka part picking) tasks. Part picking involves an employee picking a correct part from a bin and placing it in the correct slot in a cart.

One of the avenues we are exploring is building a cart that can recognize whether a part has been placed in one of its slots (and if so, which slot). The first thing that comes to mind is having some sort of proximity sensor installed in each of the slots for each cart. Keeping in mind that:

  • There can be anywhere from 15 to 50 slots depending on the type of cart (steering wheels vs glass panes)

  • These parts are extremely delicate and easily scratched (so the sensor cannot actually touch the part)

  • The fewer wires the better

  • Cost is obviously a part of our concern (1$ a piece ok, 10$ a piece, maybe not so much)

I thought this community might have some advice that could help. Any and all ideas, even unconventional, crazy ones, are welcome!

2

Are you saying that your maximum cost is $10 per sensor?

How many of these things are you planning to build?

What does the cart do if a slot is open or if all slots are filled? Does it need to communicate to something else?

Are the carts already built or are they being built as a part of this project?

Are you looking for a complete solution off the shelf or pieces to be assembled?

Are you building this yourself or looking for someone to build it?

What kind of power is available on the cart?

Do the parts have some kind of identifier attached?

3

Hi Lyndon!

We are basically developing a prototype for the manufacturing plant. If the prototype is reasonably implementable there is a chance it will be adopted throughout the plant. This means that there is theoretically no budget limit for the prototype.

However, considering that there can be anywhere from 15 to 50 slots in a cart and 8-10 people working on the floor in one warehouse, we’re looking at up to 500 sensors being needed. If the system works well, $5000 to install sensors in the carts does not seem that unreasonable… but cost is still part of the consideration for final implementation so I just thought I would mention it.

The carts are already built, and slots of different carts are generally shaped to fit the item the employee is going to be filling the cart with. To complete a batch, the employee is given a part number, part location and slot number. He/she must go to the specified location, pick out the correct part and place it in the designated cart slot number.

If the cart’s slots had sensors built into them, they may be able to communicate their state to an application we are building for Google Glass to help employees pick parts. So the workflow would look something like:

  1. Glass app directs employee to correct location

  2. App displays which part needs to be picked and what slot it belongs in

  3. Employee places part in designated slot

  4. Sensor is triggered for that slot, confirming that item was placed and allowing employee to move on to next item.

  5. repeat steps 1-4 until cart is filled (theoretically, sensors would also determine when this step has been reached)

We are not looking for a complete solution off the shelf (unless one exists…). We are open to building and assembling anything should it prove an effective solution to our problem. Depending on the complexity of what was required, we would either build it ourselves or hire help.

There is currently no power to the cart. It is on wheels and needs to remain mobile as employees push them around the warehouse to fill them with the right parts. Some of the parts do have barcodes on them, but not all, so they cannot be used as an essential part of the system.

My first thought was that we might be able to install some sort of proximity sensor in each slot. I am trying to research sensors that could be installed wirelessly but right now I’m thinking they would have to be wired to an arduino/xbee on the cart.

4

mlepic20:
There is currently no power to the cart. It is on wheels and needs to remain mobile as employees push them around the warehouse to fill them with the right parts. Some of the parts do have barcodes on them, but not all, so they cannot be used as an essential part of the system.

My first thought was that we might be able to install some sort of proximity sensor in each slot. I am trying to research sensors that could be installed wirelessly but right now I’m thinking they would have to be wired to an arduino/xbee on the cart.

Sounds like a an opto-interrupter would be the type of sensor you need. That’s basically an LED on one side and a phototransistor on the the other side. Breaking the optical path will tell you that something is present in the slot.

5

Without actually seeing the cart, slots and items it’s hard to say what would be best. However my experience suggests that no single sensor type will work unless the cart is designed around it. e.g., a photosensor as darrelg suggests may work… right up until you insert an object that happens to be transparent to infrared. Likewise, microswitches seem like a good idea, but the slot needs to be designed around the switch. Neither of those will be under $1/point.

From a hobbyist perspective it’s easy to get misled by cost: wire, sensors, connectors etc. are cheap and time is free.

From a business perspective, integrating them is far from free and you’ll likely want more expensive parts because the cost of replacing them is high. For example, I make a product that has a cable harness with 50+ wires. Even in the quantities I need to buy parts, a single wire with contacts crimped on each end is about 75 cents. Then there’s the labor cost of just plugging that wire into a cable housing and testing it. Already over $1/pin for a simple cable. If you buy a simple cheap IR photointerrupter sensor from SparkFun, then you need to have bracketry custom made to hold it in place; you need cabling, connectors, wire dressing to route the cables. By that point, you may as well buy an industrial sensor that has all that done for you at a cost of $50. Haven’t even gotten to the labor of connecting everything up.

I’m not trying to throw a damper on anything, just being realistic about true costs.

IMHO, the only way you have a snowball’s chance of hitting a sub $10/slot target is with the most dead-simple approach: a toggle switch/pushbutton at each slot and the operator presses it when he puts something in a slot. That may hit the cost target but it will have to be wired. Emphasis on may. In reality, you’ll probably end up needing a pretty rugged switch and those won’t come cheap. It is possible to build a wireless solution for each slot, but unless you’re going into the business of building these things, you probably will never recover the cost.

Sorry, but while it may be possible to prototype this for $10 per slot, I doubt a warehouse floor ready solution will be that cheap. The electronics are pretty simple: an Arduino, BeagleBone, or RPi with I/O expander will read the inputs just fine. The problem is that something that has to be pushed many times a day, that will probably have things dropped on it, get dirty over time, etc. is typically not going to be cheap.