Jauregui v. Daimler Truck North America LLC

Opinion

WO

Sandra Jauregui, No. CV-23-00729-PHX-JJT

Plaintiff, ORDER

v.

Daimler Truck North America LLC, et al.,

Defendants. At issue are Defendant PACCAR Incorporated’s (Paccar) and Defendant Bendix Commercial Vehicle Systems LLC’s (Bendix) separate motions to exclude the testimony of Tony Gioutsos (Doc. 132; Doc. 144). Each defendant filed a notice of joinder with respect to the other defendant’s motion (Doc. 146; Doc. 159). Plaintiff Sandra Jauregui filed a combined response to the two motions (Doc. 163), and Defendants filed separate replies (Doc. 170; Doc. 173). The Court finds these matters appropriate for resolution without oral argument. See LRCiv 7.2(f). For the reasons set forth below, the Court grants Defendants’ motions in part and denies them in part. I. Brief Factual Background This case arises out of a tragic highway collision in which Plaintiff’s late husband, Mr. Jauregui, drove a big-rig truck into another big-rig truck that had stalled and come to a complete standstill on the Interstate 17. Mr. Jauregui, who perished in the accident, was driving a 2022 Peterbilt Conventional 579 tractor-trailer manufactured by Paccar. Mr. Jauregui’s truck was equipped with a product manufactured by Bendix known as the Wingman Fusion, which is an Advanced Driver Assistance System (ADAS) designed to assist drivers with the avoidance of roadway hazards, including stationary objects present in traffic lanes. Plaintiff asserts that a defect in Bendix’s collision avoidance system, and/or a defect in the Peterbilt truck’s integration of Bendix’s technology, proximately caused the collision that killed Mr. Jauregui. There is no dispute between the parties regarding the nature of Bendix’s Wingman Fusion ADAS system. According to Bendix’s description thereof, which Plaintiff incorporates by reference, (see Doc. 163 at 9), “Bendix’s system included forward collision warning, automatic braking, and adaptive cruise control components which can help mitigate or prevent collisions.” (Doc. 144 at 3.) The Wingman Fusion system generates data from camera and radar sensors, from which it then determines whether objects in the vicinity of the host vehicle constitute a risk, such as the risk posed by a stationery vehicle. (Doc. 144 at 3.) The camera and radar sensors operate at different acuity levels in different physical conditions, and as a result the sensors generate a “confidence setting” that accompanies the sensors’ identification of the host vehicle’s surroundings. (Doc. 144 at 3–4.) “Bendix’s proprietary algorithm, which is incorporated into its collision mitigation system will only trigger alerts or automatic braking—depending on various data inputs—if the sensors definitively detect a stationary object or vehicle.” (Doc. 144 at 4.) The algorithm is designed to balance the competing goals of responding to potential risks and avoiding false alerts. (Doc. 144 at 4.) Thus, the proper functioning of Bendix’s system depends upon (1) the efficacy with which its sensors are able to collect data and (2) the algorithm’s processing of that data. (Doc. 144 at 4.) In support of her claims against Paccar and Bendix, Plaintiff has retained Mr. Gioutsos to serve as an expert witness regarding the alleged malfunctioning of the Wingman Fusion system during the subject accident. In motions that largely mirror one another, Bendix and Paccar argue that the testimony proffered by Mr. Gioutsos is inadmissible both because Mr. Gioutsos is unqualified to opine on the matter at hand and because his testimony is the product of unreliable methodology. Defendants also contend that the testimony proffered by Mr. Gioutsos includes impermissible legal conclusions. The Court addresses these arguments in turn. II. Legal Standard Under Federal Rule of Evidence 702, an expert may testify on the basis of “scientific, technical, or other specialized knowledge” if it “will assist the trier of fact to understand the evidence,” provided the testimony rests on “sufficient facts or data” and “reliable principles and methods,” and “the witness has reliably applied the principles and methods to the facts of the case.” Fed. R. Evid. 702(a)–(d). The trial judge acts as the “gatekeeper” of expert witness testimony by engaging in a two-part analysis. Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579, 589, 592 (1993). First, the trial judge must determine that the proposed expert witness testimony is based on scientific, technical, or other specialized knowledge. Id.; Kumho Tire Co. v. Carmichael, 526 U.S. 137, 147 (1999). Second, the trial court must ensure that the proposed testimony is relevant—that it “will assist the trier of fact to understand or determine a fact in issue.” Id. “Evidence is relevant if it has any tendency to make a fact more or less probable than it would be without the evidence and the fact is of consequence in determining the action.” Fed. R. Evid. 401. “The inquiry envisioned by Rule 702” is “a flexible one.” Daubert, 509 U.S. at 594. “The focus . . . must be solely on principles and methodology, not on the conclusions that they generate.” Id. The advisory committee notes on the 2000 amendments to Rule 702 explain that Rule 702 (as amended in response to Daubert) “is not intended to provide an excuse for an automatic challenge to the testimony of every expert.” See Kumho Tire, 526 U.S. at 152. “Vigorous cross-examination, presentation of contrary evidence, and careful instruction on the burden of proof are the traditional and appropriate means of attacking shaky but admissible evidence.” Daubert, 509 U.S. at 595 (citation omitted). III. Discussion A. Mr. Gioutsos’ Qualification as an Expert Witness Throughout their motions, in an apparent effort to downplay his technical background, Defendants repeatedly refer to Mr. Gioutsos as a salesman. That vocational moniker does nothing to diminish Mr. Gioutsos’s decades-long history as an engineer responsible for the design of numerous camera- and radar-based algorithms. (See Doc. 132-9.) Although Mr. Gioutsos’s most recent work experience focused on sales and marketing, there is no genuine dispute that he has extensive experience in the design and implementation of algorithms, including time-sensitive algorithms that integrate sensor- generated data for the purpose of executing a rapid mechanical process. Defendants argue that Mr. Gioutsos is nevertheless unqualified to opine on the functioning of Bendix’s ADAS system because he acquired automotive algorithm expertise primarily though his design of products that control the firing of airbags, not the avoidance of collisions. (See Doc. 132 at 8; Doc. 144 at 6.) Similarly, although Mr. Gioutsos has experience writing algorithms that incorporate data from both cameras and radar, (see Doc. 132-9), Defendants contend that he is unqualified to serve as an expert in this case because he has never “integrated a radar and camera system in a production vehicle” and has not “worked on brake system integration.” (See Doc. 132 at 8; Doc. 144 at 6.) Finally, Defendants argue that Mr. Gioutsos lacks meaningful expertise in “human factors,” which is the body of knowledge that concerns driver psychology, such as a driver’s willingness to accept an excessive number of hazard alerts. (See Doc. 144 at 8.) Although the Court agrees that Mr. Gioutsos is unqualified to provide expert testimony concerning the human-factor considerations that informed the calibration of Bendix’s algorithm, the Court rejects Defendants’ position that he is unqualified to opine on the algorithm itself. Defendants do not explain in what manner the sensors and algorithms that control collision avoidance differ from the sensors and algorithms that control airbag deployment. Plaintiff plausibly asserts that the two systems are similar and that, as part of the airbag algorithm design process, Mr. Gioutsos “analyzed a radar based forward collision warning system for purposes of analyzing the capability of that system to supply speed information of objects ahead to help improve the timing performance of frontal airbag crash sensing.” (Doc. 163 at 9–10.) Although a collision-avoidance mechanical system surely differs from an airbag-deployment mechanical system, Mr. Gioutsos does not profess to be an expert in vehicular braking, materials science, or any other body of knowledge related to the physical stopping of a big-rig truck. Mr. Gioutsos’s expertise encompasses only the algorithmic processing of data generated by cameras and radar, not the execution of whatever task the algorithm outputs. The Court fails to perceive why Mr. Gioutsos’s inexperience with brake systems in production vehicles would ipso facto render him unqualified to opine on the functioning of the sensors and algorithm that combine to issue a command to a brake system. Moreover, to the extent that experience with collision-avoidance algorithms is a prerequisite to opining upon such algorithms, Mr. Gioutsos possesses the requisite experience. Mr. Gioutsos has previously worked with ADAS systems that implement collision avoidance algorithms. In the early 90’s, as director of research and development for Automotive Systems Lab, Mr. Gioutsos oversaw the design of computational technology capable of “measuring the performance of ANY algorithm (e.g. All ADAS detection algorithms like AEB).” (Doc. 132-9 at 3.) Then, as a director of sales at Siemens in the 2010’s, Mr. Gioutsos developed expertise in the functioning of a tool designed to simulate the performance of various collision-avoidance technologies, including the ADAS system at issue here. (Doc. 132-9 at 2.) Defendants correctly state that “working adjacent to programmers and developers of a technology does not make one qualified to opine on the particularities of its programming and design.” (Doc. 132 at 9.) However, Mr. Gioutsos does not claim that his exposure to ADAS systems by way of his work on simulation technology by itself renders him an expert on the ADAS system at issue here. Instead, Mr. Gioutsos’s indirect experience with collision-avoidance algorithms serves to bolster his pre-existing conversance with algorithms such that that expertise may extend to ADAS technology. “The party offering expert testimony has the burden of establishing its admissibility.” Bldg. Indus. Ass’n of Wash. v. Wash. State Bldg. Code Council, 683 F.3d 1144, 1154 (9th Cir. 2012). The Court is satisfied that Plaintiff has carried her burden with respect to Mr. Gioutsos’s general qualifications. In order to defeat that conclusion, Defendants needed to provide some explanation as to how the sensors and algorithms of a collision-avoidance system differ so materially from the sensors and algorithms with which Mr. Gioutsos is familiar that his expertise in the latter is inapposite to the former, even in light of his various sources of exposure to collision-avoidance technology. Defendants have offered no such explanation. For that reason, the cases cited by Defendants are unavailing. The principal case relied upon by Defendants is Diviero v. Uniroyal Goodrich Tire Co., 919 F. Supp. 1353, 1356–57 (D. Ariz. 1996). (See Doc. 132 at 9–10; Doc. 144 at 9–10.) In Diviero, this District held that a tire engineer who only possessed experience with bias belted tires, but who lacked experience with steel belted radial tires and also lacked knowledge in chemistry, was unqualified to provide expert testimony on steel belted radial tires. 919 F. Supp. at 1356–57. Crucially, the defendant in Diviero provided two of its own experts in steel belted radial tires who persuaded the court “that steel belted radial tires are unique and complex and the experience and knowledge must have some valid connection with these tires,” “that it is important to have worked in a plant with engineers and chemists involved in control testing of steel belted radial tires,” and “that it is standard in the tire industry that one must have expertise which relates to the particular tire for the tire failure analysis.” Id. Diviero does not control the instant analysis because Defendants have declined to apprise the Court of the source of Mr. Gioutsos’s purported insufficiency as an expert. The Court concludes that Mr. Gioutsos is qualified to render expert testimony on the functioning of Bendix’s Wingman Fusion ADAS system. However, Mr. Gioutsos is not qualified to provide expert testimony regarding the human-factors considerations that informed Bendix’s calibration of its ADAS system. In his rebuttal report, Mr. Gioutsos states that “[d]rivers are accepting all of those alerts/alarms, which are quite frequently false alerts/alarms” and that “false alarms in vehicles are very common in American vehicles over the past decades.” (Doc. 144-13 at 14.) Defendants assert that Mr. Gioutsos’s expertise in algorithms does not permit him to opine on human factors. (See Doc. 144 at 8.) Mr. Gioutsos expressly concedes that those statements fall outside the scope of his expertise and are therefore inadmissible as expert testimony. (See Doc. 144-10 at 90–92.) Plaintiff does not contest this point. The Court therefore holds that Mr. Gioutsos may not testify on the subject of human factors. B. The Reliability of Mr. Gioutsos’s Expert Opinion Defendants argue that Mr. Gioutsos’s expert opinion is unreliable because it depends upon impermissible assumptions and is not predicated upon adequate data. Although the Court disagrees with some of Defendants’ reasoning, the Court nevertheless agrees with their ultimate contention. Mr. Gioutsos’s testimony is inadmissible for the purpose of explaining the cause of the accident in this case. In his expert report, after expounding upon his background and the materials he reviewed, Mr. Gioutsos describes the physical circumstances of the collision and concludes that the camera and radar sensors integrated into Bendix’s Wingman Fusion ADAS system likely had no issue detecting the stalled truck in the road. Mr. Jauregui was driving northbound on I-17 near mile post 216.1 at about 68 MPH. It was about 4:02 a.m., the weather was clear and the lighting conditions were good. The road was straight with some very slight curve to the left. The 2022 Peterbilt driven by Mr. Jauregui was in the center left lane traveling at approximately 68 mph. In the same lane was a stopped Tractor/trailer. The truck was a 2018 Freightliner. The trailer was a 2018 Hyundai refrigerated box trailer. . . . . As discussed, the night of the crash was basically perfect for a camera and an image processing algorithm (IPA) to find objects and lane lines. The Hyundai trailer had a relatively easy rear (and flashing hazard lights) for an IPA to recognize. This recognition would produce a bounding box (of the trailer) to be “fused” with radar data. The fusion algorithm would also use the lane lines to determine if the trailer was within the lane. This would make the recognition job of the camera IPA fairly straightforward in the 2022 timeframe. . . . . [O]ne can see that there are really no stationary false “objects” near the road. These would include overhead signs, bridges, tunnels, road signs, guardrails, traffic lights, poles, railroad tracks, etc. Without another vehicle (or Pedestrian, biker, etc.). A radar on the 2022 Peterbilt would have produced only limited returns from the ground, road and shrubbery other than from the Hyundai trailer. The trailer had large areas of flat metal structures for the radar microwave signal to reflect off of. For example, the rear doors and all of the metal bracketry for them. Also, the underride protection was made of metal and had large sections of flat surfaces facing the travel direction.

As discussed earlier for this crash, there was a large vehicle overlap. The impact was slightly off-center because of last second steering at the end of the crash, but this slight bias would not have affected a radar return much even in that last second. Lots of other parts of the trailer would have produced multiple radar returns to process in the signal processing portion of the radar. However, since the vehicle was not moving it would be relatively simple to group these together and determine that a stationary large vehicle was in front of the 2022 Peterbilt. (Doc. 132-12 at 9, 16–17, 20–21 (footnotes and internal citations omitted).) From there, Mr. Gioutsos assumes that the sensors were functioning properly. (Doc. 132-12 at 20 (stating that “the Bendix Wingman Advanced Fusion system’s camera and radar should have supplied adequate data for that system to process and react” (emphasis added)); Doc. 132-12 at 25 (“The failure was not because of the inadequacy of the camera, radar or computer processing capabilities.”).) Based upon the facts surrounding the collision and the assumption that the sensors were not broken or otherwise malfunctioning, Mr. Gioutsos then infers that the source of the ADAS system’s failure to issue a driver alert or engage in automatic braking must have been the algorithm’s prioritization of avoiding false alarms over avoiding collisions, as there is simply no other plausible explanation. (See Doc. 132-12 at 25 (“The Failure was because of poor decisions on algorithms and applications of those technologies in the Bendix system.”); Doc. 132-12 at 26 (“Bendix and PACCAR’s prioritizing avoidance of some false alarms over avoidance of high-speed heavy truck collisions is clearly a bad decision.”).) Defendants cite numerous cases for the proposition that a plaintiff may not infer the existence of a defect simply by virtue of an accident’s occurrence. The case relied upon by Defendants that the Court finds most apposite is Watson v. Sunbeam Corp., 816 F. Supp. 384 (D. Md. 1993). In Watson, a fire arose in a bunk bed in which a child had been using an electric blanket. Id. at 386. After the fire had been put out, a fire marshal opined that it could not be determined whether the fire had been started by a defect in the electric blanket, by the electric blanket’s misuse, or by a child’s playing with matches. Id. In assessing an expert’s assertion that the fire had been caused by a product defect, the court noted that an expert cannot infer the existence of a defect from the mere occurrence of an accident. Id. at 387–88.1 The Court views Watson and the other cases cited by Defendants as standing for the proposition that a plaintiff may not infer the existence of a defect from the occurrence of an accident where such inference is not logically supported. Thus, in Watson, the plaintiff could not conclude that the electric blanket was marred by a defect simply by virtue of the fire’s occurrence, as doing so would have improperly overlooked the plausible explanation that the fire was caused by either misuse of the blanket or by pediatric pyromania. In other words, an expert may not pick a causal explanation out of a hat full of such explanations. Here, although the portion of Mr. Gioutsos’s report described above sets forth a hypothesis that the Court might have accepted as logically sound and not in contravention of Watson or other similar cases, Mr. Gioutsos’s report later undermines the very assumption upon which its conclusion rests. Eight pages after writing that “[t]he failure was not because of the inadequacy of the camera, radar or computer processing capabilities” and that “[t]he hardware in the Bendix Wingman Advanced Fusion system is easily capable of providing the Bendix promised 3.5 second forward collision warning,” Mr. Gioutsos writes that he was “flabbergasted” to learn that the Wingman Fusion system utilizes EyeQ2 camera technology, which Mr. Gioutsos asserts is out of date. (Doc. 132-12 at 33.) He expressly recognizes that it is “possible the Mobileye [camera] IPA had trouble recognizing the box trailer with flashing hazard lights” but contends that “if EyeQ2 was the issue, it did not need be [sic]” because “Bendix should have been using the EyeQ3 chip at a minimum.” (Doc. 132-12 at 32–33.) As Plaintiff expressly concedes, the algorithm in Bendix’s ADAS system relies upon a confidence setting generated by the system’s sensors, which in turn depends upon the acuity of the sensors themselves. Without proof, or at the

1 The Court in Watson ultimately admitted the expert testimony for reasons that are not relevant here. See 816 F. Supp. at 388. very least a reasonable assumption, establishing that the sensors did indeed perceive the stalled vehicle in front of Mr. Jauregui’s truck, it is impossible to determine whether the non-occurrence of a driver alert or automatic braking was caused by a defect in Bendix’s algorithm or the interaction of its hardware with exterior roadway conditions. Throughout their motions, Defendants repeatedly fault Mr. Gioutsos for not having inspected the code of Bendix’s underlying algorithm. Indeed, Defendants assert, and Plaintiff does not gainsay, that Plaintiff never sought any aspect of Bendix’s ADAS system in discovery and indeed did not even serve a written discovery request of any kind on either Bendix or Paccar. (See, e.g., Doc. 132 at 8, 14.) The Court will not say as a matter of law that an expert must in all circumstances base his opinion of an ADAS algorithm’s functionality on the underlying code, as it may be possible for an expert to draw reliable conclusions without having done so if he controls for exogenous variables. But in this case, Mr. Gioutsos has not accounted for considerations extraneous to the algorithm, as he also did not analyze the Wingman Fusion’s radar or camera systems. Thus, much of his report is mere guesswork. Given the express uncertainty regarding whether the Wingman Fusion’s camera system perceived the stalled vehicle in this case, the only way that one could be sure the algorithm was at fault is if one is confident that the radar sensors alone generated sufficient data for the algorithm to act upon. But according to materials incorporated into Mr. Gioutsos’s report, “Stationary Vehicle Braking (SVB) is possible because the system uses BOTH radar and camera data to confirm the vehicle ahead.” (Doc. 132-12 at 22 (emphasis in original).) Based on Mr. Gioutsos’s report, the Court cannot countenance an assumption that the radar sensors by themselves generated sufficient data upon which the algorithm could have outputted a collision-avoidance command. At best, Mr. Gioutsos’s testimony establishes that one of several things might have caused Bendix’s ADAS system to not issue a driver alert or engage in automatic braking. Although such information might be helpful to the finder of fact, it runs contrary to the purpose of Mr. Gioutsos’s testimony, which Plaintiff submits to substantiate her assertion that the collision was caused by a “faulty algorithm” and “could not [have] be[en] the result of faulty cameras or radar.” (See Doc. 163 at 12 (emphasis in original).) Plaintiff may not utilize Mr. Gioutsos’s expert testimony for that purpose. Other portions of Mr. Gioutsos’s report are also unreliable. For instance, the section of his report addressing his recommended implementation of algorithmic “fuzzy logic” is of little utility with respect to determining the accident’s cause, as Mr. Gioutsos cannot say that Bendix’s algorithm does not already incorporate fuzzy logic, given that he neither inspected the Wingman Fusion’s algorithm nor so controlled for all other variables that he could logically deduce the characteristics of the algorithm. (See Doc. 132-12 at 26–30.) Similarly, Mr. Gioutsos’s implication that Bendix implemented an algorithmic shut-off function is speculation, as is clear from the verbiage of the report. (See, e.g., Doc. 132-12 at 30 (“To ‘shut the system OFF’ for high-speed differentials seemed like what happened.”).) Mr. Gioutsos’s report indicates that an older version of Bendix’s ADAS technology, the radar-only Wingman Advanced, implemented a shut-off function, but Mr. Gioutsos can only guess as to whether the Wingman Fusion does as well. (See Doc. 132-12 at 30 n.10 (“This paper was written about the Wingman [A]dvanced. But there is no reason to believe that a similar strategy is not used in the Wingman Fusion . . . .”).) In sum, Mr. Gioutsos’s conclusion that the algorithm’s calibration defectively prioritized the avoidance of false alarms over the avoidance of collisions is unreliable and therefore inadmissible. That is not to say that all of Mr. Gioutsos’s testimony is inadmissible. To the extent relevant and encompassed by his report, he may still provide useful background information on ADAS algorithms generally, and he may opine on the possible reasons that the collision in this case was not mitigated by a driver alert or automatic braking. But he may not assert that any particular feature of Bendix’s ADAS system was the proximate cause of the crash, as Plaintiff has not established by a preponderance of the evidence that Mr. Gioutsos has arrived at a reliable opinion thereon. Additionally, to the extent that Mr. Gioutsos intends to testify that Defendants’ ADAS system was defective by virtue of its use or non-use of certain hardware, he may not do so. After criticizing Defendants’ use of a camera chip from 2010 and their failure to use “dedicated short-range communications systems,” (Doc. 132-12 at 32–40), Mr. Gioutsos asserts that “[t]he cost is not really relevant in performance, because there is little to no increase in cost to make it [sic] the Wingman Advanced Fusion system capable of assisting drivers using that system in avoiding stationary vehicles in their lane when they are approaching the stationary vehicle at the 65 to 70 mph speed range” (Doc. 132-12 at 58–59.) That statement is quite literally baseless. Mr. Gioutsos does not even minimally consider the cost of developing and implementing alternative ADAS designs in his report. Again, if Plaintiff possesses independent evidence regarding the feasibility of alternative ADAS designs, then Mr. Gioutsos may provide foundational testimony addressing the technical efficacy of such alternatives to the extent that that testimony is relevant, within the scope of his report, and adequately disclosed. C. Whether Mr. Gioutsos’s Testimony Impermissibly Embraces Legal Conclusions Defendants assert that Mr. Gioutsos’s report improperly touches upon legal issues that expert witnesses may not opine on, such as legal duties, Arizona statutes, Arizona jury instructions, and the sufficiency of evidence. (Doc. 132 at 17; Doc. 144 at 13.) Plaintiff does not contest this point, but simply writes that “[d]uring Mr. Gioutsos’s testimony in this matter, Plaintiff will ensure that, while he will be asked questions about the ultimate issues in this case, none of the questions will be designed to elicit a legal conclusion.” (Doc. 163 at 17.) There is therefore no dispute regarding this issue for the Court to resolve at this juncture. IV. Conclusion Although Mr. Gioutsos is a qualified expert, his opinions are inadmissible for their intended purpose of establishing that the Wingman Fusion ADAS system’s algorithm proximately caused the collision in this case. Mr. Gioutsos’s testimony may be admissible for other purposes, such as those described supra, but Plaintiff may not submit his 1 testimony for the purpose of establishing that any particular feature of the Wingman Fusion was the proximate cause of the relevant trucking accident. IT IS THEREFORE ORDERED granting in part and denying in part Defendants’ motions to exclude the testimony of Tony Gioutsos (Doc. 132; Doc. 144). Dated this 25th day of March, 2025. CN

wefehlee— Unifga StatesDistrict Judge

-13-