CLA Series: AV Production Tip #3 – Plan for Traceability in Manufacturing Processes

Full Transcript:

Welcome to the Capital Light Assembly Podcast, brought to you by Edison Manufacturing and Engineering. Edison is your contract manufacturing partner focused on the capital-efficient assembly of complex mobility and mobility-adjacent products that are not well suited for highly automated production. I’m Brandon Bartneck, vice president and general manager here at Edison, and you’re listening to part of a special series focused on how to reliably scale the production of autonomous vehicles.

So, how can you execute and scale that production? Building autonomous vehicles is part of our core business. It’s an area where we have a lot of experience and expertise. We’re using this series as a way to share some of that knowledge with anyone who’s going through this process, so I hope you enjoy it. There are seven tips total and seven episodes here, so check out the ones before and after this, and please enjoy!

The third tip is to plan for traceability in the manufacturing process. We don't take a one size fits all approach to quality. We base our approach off the APQP process (Advanced Product Quality Planning) in the automotive world. You use failure modes and effects analyses at design and process levels (DFMEA and PFMEA), which are used to inform a control plan. This set the requirements and the plan for the production system.

There's flexibility there to figure out what are the key risks and the things that could go wrong, the likelihood of them going wrong and - if they go wrong - how big of an issue is it? One of the things that you'll see in the autonomous vehicle world is the hardware better stay on. That sounds obvious, but when safety critical sensors and components are assembled, you need to know that it’s been done correctly, and ideally you’ll have documentation to show that.

So the approach that we take is traceability. Say you're using a bolt to attach a bracket to a vehicle which has a LiDAR sensor on it. You want to know that every one of those bolts has been installed correctly. And so what you're capturing is torque value and the angle at which that's applied to assure that has been done correctly.

There's a range in which you could do this. At the minimum level, you can have a quick click gun, where you highly calibrated your calibrated gun that you're applying click. You get the right torque and you move on, and know that this has been done correctly. But a better approach, especially when you're looking at the criticality of these items from a cost and an impact perspective, is to capture and document that information - the torque and angle of information along with that unit. So what we're often deploying is something with a DC torque gun that's closed-loop and in real time is capturing the data and is putting it into our digital MES (Manufacturing Execution System), capturing and logging it alongside that product. So you have a product record that shows for each of these fasteners that have been deemed facing safety critical, you have the right countermeasure in place. And so here, torque and angle is captured and is documented. Then, when the vehicle is produced, it comes with an event finally, they get signed off and you can say, okay, this operator did this at this time, and here's proof that everything was done correctly.

I don't know if that sounds like overkill or if it sounds obvious, depending on where you come from. But, when you think about how critical some of these components are, it's a necessary step in the autonomous vehicle world. So you have to plan for that in the manufacturing process. And you need something that is adequately robust to account for that. It's one other key dimension of traceability that we often prioritize, and that is something like serial number tracking.

When you have these components, again, let's take the LiDAR sensor as an example. You want to know which LiDAR sensor is on your vehicle. When you're producing 1,000 per year and you have them driving around the road in different aging vehicles, when you have a recall upstream, you want to know which vehicles are affected. You know that by looking at serial number information for a given sensor. And it also then can be tied with software version numbers and a few other things.

But the message is that you need to be able to document that core information, capture serial numbers for key components that are installed on every vehicle and take that as input that then goes into that build book or the birth certificate that goes along with the vehicle when it's completed. There's a lot more here - this is just scratching the surface of quality - but the key tip here is plan for traceability in your manufacturing.