Latest / Data Driven with Lou Peck / Matt DiSogra and Jeff Hirsch: Bendix data and developing a translation protocol

Transcript

Matt DiSogra and Jeff Hirsch: Bendix data and developing a translation protocol · Data Driven with Lou Peck

• Lou (00:00): This is definitely going to be shorter than our average podcast. We're just talking about one topic, which is the Bendix paper that you guys just wrote. And it sounded like Jeff, you kind of reached out to Matt and got things started. Jeff (00:14): Yeah, that's correct. I'd say early last year we were talking and the topic of BDR came up and we both realized that we had a mutual interest in it and pretty much I think we realized maybe I had a piece of the puzzle. Matt had a piece of the puzzle and if we combined these efforts, this was probably actually attainable. Lou (00:39): Alright, BDR And this one, we got three people on this one, so we'll see how this goes. I'm not going to really call out whoever should answer it because you guys know best who should answer it. But for the completely uninitiated, what is BDR and what is Bendix? Matt (00:56): So BDR stands for Bendix data recorder and it's just an EDR conventional event data recorder that Bendix implemented into their ABS controllers going really as far back to around 2014, 2015 timeframe. So it's been out there for a long time. It has been historically difficult to get. Lou (01:19): Yeah, 2015. And it sounds like in reading your paper, which I found really interesting, there was some big TRB investigation that brought it to our attention in the first place, or to yours, people like you brought it to my attention. Matt (01:33): NTSB. Yeah, it was an NTSB as far as I know, and I don't want this to be like this is the official story of the origin of BDR or at least the public acknowledgement. But as far as I know, that's the first time anybody outside of, I guess Bendix had ever heard about it was there was an NTSB investigation and when that public docket was released, what was included in there was some information from the Bendix system on one of the involved trucks. And lo and behold, there was this half second by half second incremental data with speed and all kinds of other parameters that prior to no one had known about it all. And so that begged the question like, Hey, where did this come from? Where was that data hiding? And it turned out it was hiding in the ABS controller and whether Bendix was forced to I guess provide it after that point or that was maybe just the earliest point that they had it, and so then they just decided to begin providing it either way. That's around the time where you could then contact Bendix and get access to that information. Lou (02:36): Yeah, that makes me wonder what other data might be lurking beneath the surface that we don't know about. Especially on motorcycles. We got Bosch imus everywhere that we don't get data from and I've got one eyebrow up, but we get what we get and we don't get upset. So what models modules, I'm sorry, from Bendix, ABS control modules, should we be on the lookout for? Matt (02:59): So you're looking at all of the EC eighties, which is the current version, and then the prior generations called EC 60 and it's a mixed bag in the EC 60 world. I'd say probably more often than not, they don't have BDR, but there are a handful of software versions that did have it. So it's a case by case basis with the EC sixties, but given that that EC 80 rollin was around 2015, you've got well over a decade now of trucks that have been equipped with Benex ABS that have BDR. So more or less 10 years is old in truck years, so more or less everything out there is going to have BDR. Lou (03:40): We were talking a little bit about this. I haven't seen a BDR report for a long time just because I don't do a lot of heavy crashes anymore. But man, there's some good data in there that goes above and beyond what we would expect from even a passenger vehicle EDR system where we don't necessarily get brake pressure, just brake switch status. And that's one of the things that it seems like you get in A BDR report most of the time. Not of course the actual number, but in different stages. Can you talk a little bit about what data we do get on there and probably more intelligently say what I just said about the braking pressure Jeff (04:19): Focusing mainly on the EC 80. The goal of it obviously is to control DAVS, the antilock brake system. So it has the most parameters, what I would consider parameters of the recorded. You get activity about the ABS system, potentially the automatic traffic control, A TC or even ESP, electronic stability program stuff. Speed data, which is probably what everybody's most important or interested in. The EC eighties provides data, which is a really good benefit also. And then, yeah, there's some more in-depth parameters centered around brake application or really what's going on with the entire brake system when the event happened. Lou (05:08): That's huge. So you get that for, what is it, 20 seconds at two hertz? Jeff (05:13): Yeah, yeah. 20 seconds worth of data centered around a trigger, which is in the middle. So approximately 10 seconds pre trigger, 10 seconds post trigger at 20 hertz or two hertz, I'm sorry, two samples per second. Lou (05:27): That's awesome. That's a ton of data and for us in the automotive and motorcycle world, we're not used to getting post impact stuff, but that's super valuable sometimes, especially I imagine for you guys with a heavy truck where impact is very unlikely to bring it to a stop. So this helps you figure out how it got to where it got. Speaking of the, it sounds like the triggers right in the middle, what is currently triggering A BDR event? Is it similar to what we expect from other ECUs? Jeff (05:57): I would say so I think the most common one that everybody thinks about in H-V-E-D-R in the big trucks is the hard brake. There are also a few others as far as there's a collision system that's been activated, which I think bendex refers to as the wingman. There's mitigation from wingman and there is a lateral acceleration, so there's an accelerometer. If you eclipse a threshold on a lateral or side-by-side acceleration, you could potentially get a trigger. So we realize that's pretty beneficial now because yeah, you can get hard brakes, but there's also these other options for more data or more triggers or more sensitivity if you will. Lou (06:46): There's a world where you get a BDR event that's super helpful and is related to a crash, but you get nothing from, I don't know, I'm going to call it the main ECU. I'm sure you guys have better nomenclature for it, Matt (06:58): I'd say. Yeah, definitely. Jeff (07:00): Yeah, I would say so. Matt and I kind of discussed one of the bigger benefits we saw with BDR, and we may kind of agree that there might be a little bit of an educational kind of campaign about this. It's like you have a multitude of triggers and you have a wide variety of manufacturers that use this system as Bendex is more of a third party supplier to manufacturers. So you can have a bendex, ABS system and a big truck with a multitude of different engines, whether that's Cummins or Caterpillar International, paccar, some of those more difficult to retrieve data from the engine side and the ECM, whereas almost all of those will support the bendex. Lou (07:49): Yeah, that's huge. So the current method of getting the data and then interpreting it, which are two totally different things from the sound of it. Now we're starting to get into your paper. That was really helpful to give. I think everybody, including myself, a little bit of background about what the data is, why it's beneficial, what triggers it, what modules it's going to be on as far as retrieving it. I'm sure there's a few ways to do it. One, hopefully you could just go through the DLC if everything goes well, if not, then Matt, we can use all the fancy equipment in your background, which we establish is not ai, Matt (08:27): It's real, and we can use it, right? Lou (08:29): So you could pull stuff off and soldering and you got microscopes and all sorts of tools to help you do that. But in a general sense, you're going to be communicating through either the DLC or directly to the module to pull some sort of files out of the module. Matt (08:46): Bendix provides software that is able to interface with their module, it's called acom, and that a ACOM software, it gets a bunch of other things like the fall codes and configurations and things like that, but it also pulls the BDR. It just does so in a sort of raw hexadecimal format, so it's not readily readable, but the data is there, it pulls it and it saves it into a file. And so historically, if you wanted to access the interpretable side of the BDR, you wanted to use it and see numbers, then you would've had to provide that file that you downloaded through the DLC over to Bendix. You have to send it to them. They would translate it and then they would send you back the translation. So there was a bit of a process there and they had a quoted four to six week turnaround time, which is not bad in and of itself, but at the speed that sometimes attorneys want answers in cases, that's an incredibly long time to have to wait for some preliminary information about a crash. Lou (09:56): And even longer if you're trying to do some sort of testing, iterative testing where you're like, Hey, what's going to happen if we do X, Y or Z? Four to six weeks is going to put a hindrance on that. And that's really from, I read your 2025 SAE paper, which is called examination of the data structures of the Bendix data recorder for those following along, that's 2025 dash oh one dash 8 7 1 2 from SAE. And that's really what the thrust of getting this micro podcast together was, is to go over what you guys did and what you guys found there. So I guess the foundation to come into it is, all right, well, we have these BDR events, we can get the files ourselves generally speaking, or we can send it to somebody who can do chip off stuff like Matt and get the data and then typically send it off to Bendix. And this paper was to try to figure out, hey, can we, based on all of these reports that we've gotten over the years and considering who you guys are, that's a lot of them. Can we dig into this hex data and generate a reliable translation? Is that accurate? Matt (11:10): We know it's in there. We know the data's there because all you send Bendix is a singular file and you get BDR, so it has to be in there somewhere. And yeah, the question is, is there enough there that we can figure out how to translate it comprehensively, not just get one, but can we find all of the variants and permutations that they have and come up with a way to do it reliably and then be able to do it quickly. So you as the investigator can get some preliminary answers today. You can look at that data now and then you can decide where you go from there without having such a long wait time. Lou (11:50): Yeah, and like you said, you have a bunch of different, obviously there's a ton data that comes out, so just translating that sounds like a nightmare. I was just talking to Ed Inger about what he did with the Kawasaki stuff, and it sounds like this is probably a similar process. Sounds like some of the groundwork you at least had some headway from Jeremy Daley and his crew, but far from complete. And then you have all the variants, you have the EEC 60, you have the EC 80, both of those have different variants. So what was the process? How did you go through generating those translations and getting to a place where you're comfortable? Matt (12:30): It was really just a data assembly problem at that point. We had to gather up all of the examples we had over the years that we've gotten translations from the manufacturer and then kind of lay out and say, okay, well do we have enough of each style? Some are more common than the others, and do we have enough of every single one to be able to explore all the different permutations There could be in the sense of a particular value could have, or a particular data element could have a value from zero through seven. So if we want to know what the raw data looks like, we need to make sure we have an example that's got a zero, a one, a two, a three, a four, a five, six, because we need to make sure you could see some logical patterns, but you want to be sure. So that was probably the hardest part was digging for enough examples where we knew we had validated every single combination there could be. That was the tricky bit. Geez. Lou (13:25): Okay, now I'm starting to wonder how long this project took one, and then how many data sets did you need or did you end up analyzing to accomplish that? Matt (13:35): What was that number, Jeff? Like 430 something? Jeff (13:39): Yeah, 438 individual events or records across 86 different reports where, so you could say basically we analyzed 86 individual ECUs or memory files from 86 ECUs, which was a combination of 438 events. Matt (14:00): And you're searching for the weird stuff. Lou (14:01): Yeah, you're searching for the weird stuff. And then I imagine there's some single point data, but most of it is two hertz, 40 samples each channel and a bunch of obviously different potential values for each one of those channels. So that's a lot to tackle. Matt (14:18): It is, and the trickiest part might really have been that initially you said, okay, you've got 40 individual samples, you've got 22 channels, so you've got X number of bytes if we're just doing one byte per value or something like that. And that was the first way we attempted the problem. And when you look at the raw data, it's like an order of magnitude too small to be using that. So immediately we knew they're cramming multiple values into a single bite. So they're doing a binary level conversion, which is even more challenging. You have to kind of unpack each bite, whereas in other systems, like other heavy vehicle, we see a one byte per value convention, they're not trying to cram stuff in, but it would appear for probably the sake of memory savings. There's lots of data condensed into a single bite, multiple parameters. Lou (15:11): One of the most interesting things to me was that the DTC report that comes out, hopefully I'm using that right, correct me if I'm not comes out as A PDF and then you get an HTML, but the HTML file, the data is not readily visible because all of the data that we actually care about is hidden as a comment. Super interesting. So you're digging in there, that must've been kind of a eureka moment when you pulled up that HTML and were like, oh, there's all the data. I just removed the asterisk or whatever as a comment indicator in HTML and now I have all this hex data. Granted, it doesn't help me figure out the vehicle speed yet, but that's where the data is. Jeff (15:48): And I guess to add to that, that was step one. Maybe we discovered, okay, here's this huge blob of data if you want to refer to it as that. So to you and I and most people, the human eye, you read it and it just looks like gibberish and hexa decimal. And so probably one of the biggest first undertakings was trying to structure that. We got to tape this and look at it from the aspect of, to me it was like the analogy of a puzzle. There's pieces in here that come together to make this whole data file. And it was talk about more the methodology of where we started. Okay, well how big is the whole data block and what makes sense rationally? And we knew well, it's got to be broken down into sections. There's actual data for the event. There's an event header for each event that lists parameters about engine hours and timing and what the trigger was. So there's a header for each event, there's the actual event data, and then there's another section of the file that gives an overall system information. And so I looked at it as we're going to take this one blob of data and we have to segregate it down in these chunks and identify where does one start, where does the other one end? And can we repeat that with confidence? Lou (17:14): And that was one of the cool things to me in this paper is that if I read it correctly, at the end of the day, you had a hundred percent accuracy in interpreting those HTML files or the hex data is probably a better way to say it. Matt (17:28): In the sample set we had when we published the paper, we didn't find any examples where our method didn't produce a file that matched exactly what Bendix had provided. And it's a pretty big sample set. But by no means would I say it's comprehensive. It's now been quite some time since we put the paper out. We've done substantially more translations since then and gotten some Bendix translations. And so far I have not found any problems with the method in the paper. No one has come to me and said, Hey, because some other folks out in the industry now that have got this paper have implemented that processed themselves and they're translating their own reports. And that was kind of the whole spirit of the paper. And I haven't gotten any emails saying, Hey, we found an issue and I don't want to guarantee that it's totally error free. But so far it seems, I think we caught in our sample set all of the edge cases. That was the whole idea is catch every little weird possible bit combination. And I think we got it. Lou (18:26): This episode is brought to you by light point of which I am the principal engineer. Light Point provides the collision reconstruction community with data and education to facilitate and elevate analysis. Our most popular product is our exemplar vehicle point clouds. If you've ever needed to track down an exemplar, it takes hours of searching for the perfect model, awkward conversations with dealers, and usually a little cache to grease the wheels then back at the office. It takes a couple more hours to stitch and clean the data. Time is the most valuable thing a person can spend, don't waste yours doing work that's already been done. That Light Point has a huge database of exemplar vehicles all measured with a top of the line scanner Leicas RTC 360. So no one in the community has to do it again. The exemplar Point Cloud is delivered in PTS format, includes the interior and is fully cleaned and ready to drop into your favorite programs such as Cloud Compare three DS Max Rhino Virtual Crash PC Crash among others. (19:23): Head over to light point data.com/datadriven to check out the database and receive 15% off your first order. That's light point data.com/another. One of the big benefits of being in a position like you guys are where people know you, you're the authors on this paper, they're going to reach out to you with weird anomalies, which I see on the motorcycle side all the time. And it's super helpful to have that hub, that repository of all that information where you can then disseminate it to future well to other practitioners and incorporate it into future research. So that's impressive to hear. How long did this take you? This seems like a giant project. How many people were involved total and how long did it take you? Matt (20:09): I forget when Jeff and I had put our heads together and kind of realized that, like he said, he had pieces of it and I had some pieces of it. And if we put our pieces together, we were like, oh, I think we have the whole thing here. The first issue really was that we were like, well, what do we do with this? Publishing was not our first idea because maybe intuitively you think like, oh, maybe I'll start being Bendix Jr. Or doing translations or something like that. But I think if you subject that to any kind of analysis, you quickly realize where's all this stuff going? It's going into court and you can't roll into court with some secret proprietary translation that you won't tell anybody how you do. But it's essentially a trust me bro approach, which it's never going to work, so there's no way to do it. (20:58): So then you sit and well, all right, you think to yourself, what do we do? The only answer really is if people want to use it, we have to validate it and to validate it, we have to show the method. We got to put it out there in its entirety. So that took a while maybe to get to probably maybe, I don't know, a month or two of going around and around of what do we do with this? And when we settled on publishing, then we looked up and we said, well, all right, when's the deadline for SAE papers for World Congress? And we're like, oh. Lou (21:27): And it's always a week away from whenever you look it up, by the way. Matt (21:30): I think it was like two months away or something, which is close. It was not fun. So we got to work heavily and quickly and got in for the SAE deadline. But yeah, so it was, I don't want to say chaotic or frantic. We were cool and collected the whole time, but of course it was busy. It was a busy two month block there. Jeff (21:51): Yeah, I think, what was that? Probably springtime last year when we started really talking about it. And I think the majority of it really started to come over or come together throughout the month of the summer is when we really got down to the core of it and realized, yeah, we can make the deadline and then start writing up the draft to publish it. Lou (22:15): So it sounds like number one, practitioners need to be on the lookout for any heavy units with Bendix ABS systems. Currently 2025 is essentially every Bendix unit that's on the road currently going to be an EC 60 or 80. Matt (22:36): They're all going to be eighties and they're going to be in every truck on the road currently except for Daimler products, which would be Freightliner and Western Star. But otherwise, everything else by default is going to be Bendix. I mean, you can, of course, if you're big enough fleet, you can order a truck with whatever you want, but the default option for every other OEM is bendix. So it's very common. Lou (22:57): Dang. And the other one's going to be Wabco, Waco, Matt (23:00): Which so far we don't know anything about the wabco, but even in all of the trucks that have the Bendix, if you take a step back and look at the state of heavy vehicle EDRs presently, you take trucks rolling off the line today or last year, a lot of the default eds, what you call the engine based EDR or the O-E-M-H-V-E-D-R have problems or are just maybe not good enough for the kind of calculations you want to do in a reconstruction. They might have timing anomalies. We know about some things there. They might just be non-functional. Some OEMs and manufacturers that say they have an EDR and they, as we all know, you hook up and it turns out don't, a lot of them are only one hertz, which is better than nothing, but it's limiting for a lot of analysis. So I think when you look at the BDR, it's a better data recorder by almost every measure, you get more parameters, you get a higher sample rate, and you're getting speeds from the wheel ends instead of the transmission output, which is the common convention for everybody else. So you're getting wheel speeds from an independent source. So it's not just to your question what ones might have it, it's more like I'm excited knowing that the BDR is available now because I know I have a really good data recorder as my, that's like my plan A, and then the OEM heavy vehicle is almost my plan B because it's not as good by almost every measure. Lou (24:29): Yeah, you're not going to get that. The brake pressure, you're not going to get steering, you don't get steering angle, Matt (24:34): Nope. Lou (24:34): You'll get speed, but not to the same resolution. That's awesome. So that's kind of the first, yeah, like you said, that's your priority is getting the Bendix data. So it sounds like based now with this paper coming out came out four months ago, practitioners have three options. One, they can read the paper, implement the methodology themselves themselves once they get that BDR file and the HTML and look in, find the hex data, do the binary conversions where necessary, turn that into real world vehicle data. They can, I think, send it to Bendix and get the interpretation wait four to six weeks, or they could send it to either of you guys. Is that a fair summary? Matt (25:18): That is, and I think that's how a number of folks have implemented it themselves is you can do this in Excel, for example, and they've implemented it and they run the calculations and they get their answer the day they're out at the truck and then they still send it to Bendix to get their copy, but they can at least give their clients answers. They have high confidence in that day, but if they feel comfortable going into court with the manufacturer translation, they're still doing that. So they're doing both option one and option two, and then of course, yeah, we're able to do it. Obviously we did the paper. I have the ability to translate these, so there are some folks that don't want to do it themselves and we're happy to provide that for the ones that need it, Lou (25:57): I'd firmly be in that camp. There's zero chance, Matt (26:01): Not a hex guy. No. Lou (26:03): The data, I read the paper and I was like, that's a lot of work. And yeah, I'm also not a hex guy. I do enjoy coding from time to time, but this is not a place where I would feel like my time was well spent and I would come up with a good answer. So I'm going to read real quick a summary, just a quick summary of what you guys put in the paper at the very end where there was two major breakthroughs. One, you can perform independent analysis of the raw BDR data in conjunction with the manufacturer's translation. And then two, you can perform analysis without, should that ever be needed. Yeah. Okay. And then like you said, this newfound ability, not super newfound, but pretty darn, it sounds like you guys have maybe had some sort of close capability, some capability of this for a little bit of time within the past year or so. But that allows you, like you said, Matt, to very quickly let anybody involved in the case know what was happening with the truck and not have to wait that four to six weeks before they get any feedback as to what might've happened in the crash Matt (27:17): Or maybe the system didn't capture the event we're interested in, and then we can save ourselves that time and effort of sending it off just to get a record back, that's not useful. So we know every time f we're going into sending it in for the manufacturer translation, we know what we're expecting to get back and we know it's going to be useful. So it it's a worthwhile effort there. Lou (27:39): And that was a hex joke by the way. The f's, that wasn't a swearing joke. Matt (27:43): I know what you're saying Jeff (27:43): For those that dunno. Yeah. (27:46): But yeah, to add to that, I think the word that comes to mind focused around the paper and for everybody else is accessibility. The goal is accessibility. I think we're all pretty familiar with the field work side of it. Like we talked about, we know how to get the file, we know the data's in the file, how do we do something with it in more of a timely manner to at least get some kind of an insight. And that was the real takeaway from this. If we can provide that accessibility, it's just that much more beneficial. Lou (28:20): Yeah, that's awesome. I applaud the efforts and the amount of work that it must've taken just seems mind boggling. So hats off and what do you guys have your sights set on next? You're going to tackle Waco or is that not common enough to require such an effort? Matt (28:39): So there is the raw data that comes out of Waco's and peeking at it, it doesn't appear they're doing anything too sneaky. They're not hiding anything in there. So I don't think there's additional data in Waco. I think what I hope to see and what we're playing around with, but what I really hope other folks might take this research and do with it is more testing on Bendix systems, more validation, better understanding how they perform. So now that anybody out there can access the BDR, do some instrumented testing, controlled testing and get that feedback in real time, hopefully other research is now much easier. So I'm looking forward to seeing what folks use it for in terms of doing more testing on Bendix systems and to help us collectively better understand how to use them in our analysis, what kind of limitations there might be, that sort of thing. Jeff (29:33): Well, yeah, and not only that, I would like to see from this, like I said, I used the word educational campaign. Maybe we can get a little more of an understanding, hey, there could be something in the ABS controller, just don't overlook it. We were talking a little bit about manufacturers and I just had a thought come to mind. Matt mentioned Freightliner with Waco, but then I remember there's always these caveats or anomalies. I see a lot of Freightliner M twos in the medium duty sector that have a Cummins and their bendex. We were pleasantly surprised last fall to run across an Isuzu NPR model kind of cab type truck that had a Cummins, it had a bendex, right? And so I would hope more people give it at least a second thought like, Hey, we need to make sure that we don't kind of glance over this going forward because there's such a wide variety of applications where we're finding this, Lou (30:39): And I was telling you guys on the call, the prep call, I was at an inspection, it was a couple years ago, times change, people evolve, but the reconstructionists on the other side, I was asking him, Hey, are you going to download the BDR? And he is like, no, I haven't really found anything fruitful from those. And that must just have been a little bit of not knowing exactly what was in there or how to get it or hadn't gotten it back from Bendix before. But to your point, Jeff, that's the first step in all of this data stuff is knowing that it's there and that it should be pursued. So that's awesome. And then to your point, Matt, is now we're in a much better position to be able to validate and test fringe situations because you can be there with a laptop and get the file and interpret it right away and say, oh, interesting. When I do x, Y or Z, this is what the data looks like. So if we see that anomaly in a real world case, now we know what to do with it. So that'd be cool to see an influx of papers on BDR as a result of this capability of rapid testing now. Matt (31:43): Yeah. Yeah. Jeff actually brought up a really good point there about the medium duty world, right? Because the medium duty world lives in this no man's land of, you've got vehicles that are too large for 5 63 to apply and they're too small to get the big diesel engines that have the heavy vehicle, even though those aren't even required. So there's nothing, there's no requirements and they don't get the benefit of something that has it anyway. And so I'm sure everybody who's been out on all these type of medium duty vehicles knows, yeah, you plug in, poke around for a little while, but there's really nothing useful. And that's what we're seeing is some of the larger end of the medium duties have Bendix pneumatic brakes and they have EC eighties and they've got a unique software in them, but they have BDR. So it's totally unlocked the medium duty world again in terms of useful EDR coming off the vehicles that otherwise there'd be nothing. Lou (32:41): That's huge. So can you interrogate the BDR just through the DLC or OBD two or whatever the medium duty trucks have? Matt (32:48): Yep. You just treat it like any other heavy vehicle and it works exactly the same. And if you've built the translator, your own translator from the paper, the methods apply a hundred percent. Lou (32:59): That's huge. So this might be a naive question. I don't do a lot of work on the medium duty stuff or any really come up now and again, but I would probably email or call you guys to help. What about Ford F five fifties where they don't have an airbag, they do they ever have bendex stuff? Matt (33:18): You've got to be in the territory of pneumatic brakes. That's the change Lou (33:23): That makes Matt (33:23): Sense. Lot of the small mediums are a lot of times hydraulic ABS and then the big mediums are pneumatic. So you've got to at least be to pneumatic. I've not, I don't think the EEC 80 can drive hydraulic ABS. It's pneumatic only. Lou (33:40): That makes sense. That's the world Bendix lives in is pneumatic breaks. Alright, those are all the questions I've got. Is there anything that I didn't cover that you think is worth mentioning about this work or how it affects the community? Matt (33:55): I think that's probably it. There's the thing that I could say that you can decide whether you include this in some way or another. I think the state of the industry as a whole may be moving in a direction where there's a lot of data on a vehicle and a lot of it is not readily accessible. And so techniques and methods for getting that data are going to become more important and folks who are interested in that, it's just we're going to need people who want to go and explore and try to find that information because there's useful data there. But there is seemingly a trend of manufacturers wanting to not make it easy to get the data despite it being there and being useful. That just might be a direction that the industry is headed. Lou (34:46): Yeah, we need pioneers and people who are interested in figuring out how to get data that were not necessarily being told exists. To your point earlier, or actually it wasn't your point, you just mentioned it. We didn't find out about this bendex data at all until there was some large governmental investigation into this catastrophic event, which made me think, has this been there before? And they just weren't telling us. You might know the answer to that, but in any case, like I said, there's motorcycle is a lot of kms and a lot of Ducatis have Bosch IUs, which we know Bosch as the developer of the CDR tool among other things. But it just seems to me like there's probably some sort of diagnostic data under there. If you have the right tool and the time to look into it and implementing some of the methodologies that you guys are using, or at least the philosophy and the concepts. Maybe we can find some additional data that doesn't currently, that isn't currently known to the community that so young people or old anybody who's got the energy is like the whole community needs that kind of rigorous investigation. Jeff (36:07): Yeah. Well, to Matt's point though, kind of like our thought was we were fairly confident. We knew it was there, but it was still, let's take a dive in and see, and then you look big picture like what Matt said, it's like, well, this has been around 10 years, so it took a little bit of time. But to Matt's other point, what's most exciting to me going forward is as technology advances, your microchip technology gets better. You have more room for memory, you have more data being transmitted on the bus throughout the vehicle, you have more of an ability to pick up parameters. And it's kind of exciting to think about maybe a little bit of a Pandora's box, but it's like there could be something there. Right now it's unknown, but it's a mystery. But can it be solved and is it worth pursuing that Lou (37:05): Always, I just wrote one of my weekly newsletters and I put in this quote that I love, and it's just without data, you're just another person with an opinion by, I think it's William Edwards Denning and such a great quote. So it's just like I put a hundred percent of effort into finding any data that might exist on a machine that's involved in a crash. I'm investigating diagnostics. I'm still a little bit on the fence about diagnostics for motorcycles. So I still have, I think there's so much work to be done there to what you're talking about, Matt, if anybody in the industry wants to put a keener eye on motorcycle diagnostics, there's a lot of work to be done there. One of the hurdles there is exactly what you guys just jumped over, which is if you want to get a full diagnostic report on most motorcycles, you're going to need to be a dealer, have a dealer license, and we can't get that. (38:06): So yeah, there's challenges there, but I think there's a lot of work to be done there, and maybe we can find something fruitful and we haven't really found much fruitful yet, but little yeah, tangent at the end there on my part. I didn't mean to hijack things, but thank you guys so much. I really appreciate you spending the time. I think this is really valuable research and really, like I said, just hats off on the amount of work that you put into this and the a hundred percent translation success that you had at the end of the day. That must have felt amazing. So thank you guys for taking the time to share your knowledge. Matt, I think you're the first repeat data-driven podcast guest ever, so thanks for coming back. Matt (38:53): Oh, I've had a lot of titles put on me over this last year, and I'll take that one. That's a lot better than some of the other things I've been called. Lou (39:04): Yeah, and thanks, Jeff. I really appreciate it, man. It's great to meet you. I mean, we haven't met before this. I've seen your name around, but it's been a pleasure and I really appreciate your time. Jeff (39:14): Yeah, I agree. Happy to be here. Happy to share more importantly, happy to talk about this stuff. It's very interesting, I think to Matt and I both, and so you kind of asked earlier, I would say there's definitely more areas going forward that are intriguing. There's more rocks or stones to turn over, more places to look and more potential. We'll just see what goes, how it goes in the Lou (39:41): Future. Oh my gosh, yeah. With the ADAS stuff, both on the heavy side and the passenger vehicle side, this is going to be a wild decade coming up. Oh yeah. Oh yeah. Thanks again, guys. Hey everyone. One more thing before you get back to business, and that is my weekly bite-sized email to the point would you like to get an email from me every Friday discussing a single tool, paper method, or update in the community. Past topics have covered Toyota's vehicle control history, including a coverage chart, adas, that's advanced driver assistance systems, Tesla vehicle data reports, free video analysis tools and handheld scanners. If that sounds enjoyable and useful, head to light point data.com/to the point to get the very next one.