Here is the transcript of the webinar, "Understanding the Benefits of Connected & Automated Vehicles and System Controls for SMART Mobility," presented by the U.S. Department of Energy's Vehicle Technologies Office.
.....
David Anderson:
I will remind you all that this Webex call is being recorded and it may be posted on DOE’s website or used internally if you do not wish to have your voice recorded please do not speak during the call. If you do not wish to have your image recorded please turn off your camera or participate by phone. If you speak during the call or use a video connection, you are presumed to consent to recording and use of your voice or image.
OK, good afternoon, everyone. I’m very happy to welcome you all to our fourth in a series of webinars to discuss the results, the insights and the conclusions, from our first phase of the DOE Smart Mobility National Lab Consortium. My name is David Anderson. I’m the program manager for EEMS, or Energy Efficient Mobility Systems. That's a program of the Vehicle Technologies Office, part of the Department of Energy's Office of Energy Efficiency and Renewable Energy. So, we've been having these webinars every two weeks each one focusing on a different aspect of what we accomplished and what we learned through the Smart Mobility consortium's research over the last three to four years. Our first three webinars were really focused on our large-scale transportation system modeling workflow and the mobility energy productivity metric that we used to measure the energy the cost and the time of regional scale mobility systems, but today we're going to focus on a specific aspect of future mobility that is connected and automated vehicles or CAVS and the potential benefits that they may provide.
Eric Rask led the CAVS research pillar of Smart Mobility while at Argonne National Lab, and he's going to be the primary speaker today. But before I turn it over to Eric to give you some of the highlights from the Smart Mobility's CAVS research, I’ll again give you a brief background on the consortium. So EEMS is one of several research programs within DOE’s Vehicle Technologies Office or VTO. VTO conducts research and development of advanced vehicle and transportation technologies that improve the efficiency of our vehicle fleet and of our overall mobility system VTO supports component level research on advanced combustion engines and fuels advanced materials energy storage and electric drive systems this research is done with a focus on improving the energy efficiency of individual vehicles the I’m program brings the system level purview. To VTO from a technology perspective that means understanding how connected and automated technologies impact not only individual vehicles and the powertrains in them but researching technologies that improve system level efficiency at the intersection and traffic network level and ultimately working across all passenger and freight modes to develop solutions that improve mobility at the entire urban or metropolitan scale.
As I’ve said before, because the opportunity space for transportation is so large and complex, we've taken a multi-disciplinary consortium approach to mobility. We convened five of our leading national labs doing mobility research to build a multi-year multi-laboratory collaborative dedicated to further understanding the energy implications and opportunities of advanced mobility solutions and we call this the Smart Mobility National Lab Consortium this is really a key research effort of the I’m program and is focused on understanding the system level impacts that emerging transportation technologies and services will have on mobility and on identifying solutions that improve mobility energy productivity.
The consortium was designed to consider transportation holistically that is as a system a highly interactive highly dependent very complex system of systems we did research to understand each of these systems individually. But more importantly we work to understand how they fit together, how they interact and how they affect one another. Today we're going to focus on two areas shown here. Connected vehicles and automated vehicles. Now we often lock these two together because certain types of automated applications are enabled or can be improved by connectivity but to be clear connectivity and automation are two separate technologies. Connectivity refers to communication an exchange of information between vehicles. That is V to V or between vehicles and infrastructure that is D to I connectivity can apply to both human driven vehicles and automated vehicles. Now automation refers to some part of the driving task being handled by a non-human control system a computer controlling your speed acceleration braking or steering for example again automation can be enabled by connectivity in which the control uses information provided to it by V to V, or D to I or we can have automation without connectivity in which the vehicle relies on sensing technologies such as lidar and radar and cameras to sense its environment and make control decisions. But again, these two technology pathways are often merged in their applications, so we refer to that as CAVS in the shorthand.
So, what is a CAV? Well they range from current OEM developmental vehicles and prototypes like those on the top here, to research lab vehicles and concepts like those at the bottom. CAVS are one of the primary technology levers that needs and Smart Mobility considers when looking at the future of transportation and while nearly every new vehicle model sold in north America today currently has automated features as either standard or optional equipment, I think adaptive cruise control or lane centering or automatic braking. We don't yet have fully automated vehicles ubiquitously driving everywhere on our streets. So is that in our future we ask ourselves what degree future mobility will be automated. Some forecasts like this one that I showed in our last webinar predict that the share of automated vehicles will grow a lot in the coming decades with a third of vehicle miles traveled being served by automated vehicle services and half served by personally owned automated vehicles. So as new automated driving features are introduced we have to understand how they operate not only how they impact the vehicles equipped with them but also how they affect the surrounding traffic and that's what our CAVS research building set out to do and what Eric will describe in just a moment.
So back before we started the Smart Mobility consortium, we set out to estimate what are the potential bounds of energy outcomes that could result from the large-scale deployment of connected and automated vehicles. So that led to the study that you see here by researchers at NREL Argonne and Oakridge and I’m a little hesitant to highlight the study but it’s been cited many times since 2016 so I want to mention it as a starting point for Smart Mobility. This analysis of existing research at the time concluded that in the worst case we could see a 200 percent increase that's a tripling of energy consumption from connected and automated vehicles. So that's the upper bound estimated by looking at potential impacts in both the travel demand side and the vehicle energy efficiency side the increase here is primarily driven by induced travel that's an increase in VMT due to the EEMS and convenience of fully automated travel.
In the other hand this analysis showed that if we use connectivity and automation to improve the efficiency of our transportation system. We could realize up to a 60 reduction in transportation energy again driven by both changes in efficiency and travel demand.
The primary drivers here were shown to be opportunities to right-size vehicles into smooth drive cycles and operate more efficiently. So, what this research told us was that there's a lot of uncertainty and we saw a big gap in understanding the energy impact of CAVS. So, the answer to what are the energy impacts of CAV technologies came out to be it depends to elaborate on that the impacts of connectivity and automation depend on a lot of things levels of automation. So, most of you are probably aware of the SAE automation levels zero through five. The impact of any technology across the spectrum varies depending on what it is how it’s implemented and how it’s used types of connectivity you can achieve very different results in terms of traffic flow and energy consumption, depending on the information that is communicated to or by the vehicle classes of vehicle a particular automated feature can result in different energy impacts if it’s on a compact car versus a full-size SUV or even a medium or heavy-duty truck. Similarly, whether the vehicle is powered by an internal combustion engine or a battery an electric machine or a combination of the two in a hybrid powertrain will determine the energy impact of the CAV technology.
What's the controls objective is the feature focused on safety for driver comfort and convenience. Is it traffic flow? Or is it fuel efficiency? Sometimes these objectives are in opposition to one another, but often they're synergistic technology adoption the impacts of 10 penetration of a particular CAV technology can be very different than if half the cars on the road are equipped with it and the relationship is not always linear.
Operating environment is the cap being used on the freeway or on or on an arterial road are there signalized intersections are there curves? Or hills? Right is the traffic heavy or light? All of these factors affect performance and finally test procedures. If we want to show empirically what the benefits of CAV systems are or if we want to validate our models that give us a simulated result how do we do that in the laboratory or on a test track or in the real world calf testing is an emerging research area that's actually very complicated.
So now that I’ve teed up a lot of difficult questions, I’ll turn it over to Eric Rask who led the Smart Mobility research pillar at Argonne National Lab to try and provide some answers. Eric will give some highlights of our CAV focus research explaining some of our findings and results and while I believe we made a lot of progress in this area over the past few years. I think it’s clear that there's still a lot of work to do not only by DOE but by the entire research community. Now once Eric is done, we'll have a few minutes for Q&A. I encourage you all to type your questions in the Q&A box not the chat box but the Q&A box on your screen at any time and we'll go through as many of your questions as time allows so now I will turn it over to Eric Rask.
Eric Rask:
Hi, everyone! Hi, thanks to David for introducing our topics and then I’m looking forward to giving an overview of some of the highlighted research areas as well as some of the thinking behind what we did. So again, really building within the smart consortium and now we're moving forward so looking forward to the discussion.
So starting out really at the highest level we're talking about what kind of research questions are we thinking about and as David mentioned we're really trying to understand energy technology and usage implications of connectivity and automation and then we're also trying to think of efficient connectivity and automation enabled solutions. So, the first part you can think about some of the questions related to VTO's technology portfolio and how do we identify synergies both positive and negative with connectivity and automation relative to electric machines. Batteries all these other technologies that we're interested in and then the second part of this is how do we really identify areas where connectivity and automation can be harnessed. As an opportunity and an enabler for improved energy efficient better traffic flow less congestion and all of the other items that v2x technologies are considering specifically we're trying to start then breaking into three specific research questions. One how will be connected, and automated vehicles and systems behave in the real world? So, thinking a lot about what really will happen when these vehicles are introduced and how they behave two what are the energy technology and usage implications of connectivity automation and the combination of both technologies. Really highlighting some of David's introductory points we want to understand about connectivity automation and then how do we combine them synergistically. And then lastly of course is what are the best ways to harness connectivity automation for reduced energy consumption and improve mobility and transportation.
So, beginning with those three foundational research questions we sort of break that into research focus areas. So, the first of which is prototype development experimentation and large-scale data analysis. So, if you think about our initial research question how do these systems work in the real world, it’s truly an experimental prototyping and partnering with large data sets and understanding how these vehicles are operating. Moving into the second research focus area we're really interested in how do we evolve and validate particularly using the data from focus area one how do we evolve and validate our connected vehicle automated and connected an automated vehicle modeling portfolio. Certainly, mentioning this is integrated within the Smart Mobility modeling workflow presentations and then ultimately, it’s into this larger effort of how we work from the vehicle level all the way to the system level. A third research question really understanding the impacts of connected and automated vehicles within the near-term transportation system so this is specifically thinking about what happens when different penetration levels of these vehicles show up on existing roadways. Existing transportation systems there may be some degree of communication vehicle to vehicle or vehicle to infrastructure, but we aren't necessarily harnessing all of the capabilities afforded to us and then of course building on that team the fourth question. Is CAV enabled opportunities for reduced consumption and congestion? So really understanding what's possible from the range of these different technologies.
So now we're kind of getting into some research highlights and I really want to stress that we're focusing on research highlights. Here there's a really immense amount of exciting and really innovative work within the connected automated vehicle focus area as well as the smart consortium in general. So I’m really just starting to talk about a few highlighted research pieces and then maybe pulling back one layer of the onion to sort of really start understanding how did we get to these numbers and some of the background insights from these pieces so one of the large-scale experimental work pieces that we did was looking at efficiency opportunities for class 7 and 8 trucks. So specifically we really started looking into truck platooning and the really interesting high level conclusion here we're looking at about a six to an eight percent fuel consumption reduction for overall class seven and eight fuel savings and so if you think about that that's over a billion gallons saved annually of diesel fuel. So once we start thinking about how we get to those benefits we break that into two separate pieces one of which is the efficiency benefits available because of truck platooning and then the second part of that is the opportunity to actually engage and observe those benefits through the availability of platooning.
So, stepping in again. Stepping in one layer the first part is we see a comprehensive study from the Department of Energy so these are some pictures from actual testing we really did a lot of very detailed experimental work looking at different truck platoon configurations. Different following distances and that's really how we get the first piece of this number in terms of seven to thirteen percent savings depending on how many trucks are in the platoon. What the configuration is and what type of speed they're operating at breaking that now maybe into one other layer we have some interesting results about trucks on the individual level during these platoons. So the figure on the left shows the breakdown from the lead middle and trailing vehicle and so if we look at the chart it’s really interesting to see that as the separation distance decrease for these vehicles we see a dramatic increase in fuel savings for the lead in the middle vehicle we see a similar trend at larger separation distance. For the trailing vehicle but we actually see the benefits start to decrease as the separation distance gets very small and so while I’m not going to talk much about that here a lot of really interesting aerodynamic work was done here both in terms of instrumentation and data analysis to really understand some of these very interesting and not necessarily expected trends for this trailing vehicle reduction as it gets closer and closer to the other vehicles the figure on the right again highlights some of the comprehensive nature of this analysis. Not only did we look at two and three truck platoons we looked at platoons that may or may not be following an SUV we looked at vehicles that have cutting in and out of the platoon so really trying to understand not only the pure benefits of platooning operation but trying to get some insights about the real world sensitivities and some of the real world operating conditions that may be encountered impacting the overall benefit.
Again, we talk about the benefits of platooning and now we're talking about the opportunities for pursuing. So this was really a set of studies to understand what type of platooning capabilities and opportunities exist at a national level so in this study we worked with about 57,000 vehicles provided by Volvo as a data partner for this study and so we started to talk about how often are these vehicles driving over 50 miles an hour.
So seeing a platoon benefit similarly we then wanted to identify locations and trucks that had a platoon formation capability and that was defined in this study as about a 15 mile radius of truck to truck as well as a 15 window of travel time. So really understanding how close and available these trucks are to form platoons that can then drive on long interstate pieces and again circling to the top conclusion we see about a 60 percent opportunity for platoonable miles. So combining those two numbers together that's how we get the six to eight overall benefit in terms of opportunity to platoon as well as the benefits of platooning itself so a really interesting study a lot more detail available but really highlighting some of these high-level conclusions next up another experimental question.
So, David talked about automation and really trying to understand the functionality and electrical loads associated with connectivity and automation. So, starting maybe from some of the early field trials we saw a lot of vehicles with very high electrical loads associated with automation and the compute required for some of these operations. So something like maybe 2000 watts required for driverless draw capability so being able to do all of the sensing and all of the awareness and all the operations to be a fully driverless vehicle the second part that we wanted to understand is what does. The spectrum of capabilities looks like because we're interested not only in fully driverless capability, but we're again interested in automation as an enabler across the spectrum of capability. So, we really wanted to understand things like highway cruising eco approach and signal departure. Highway lane changes and then just low speed repositioning and low speed operation and so field studies here make us pretty confident that we're talking a much lower number maybe 400 watts for automated eco driving and then in the next slide we'll actually talk we see other operations that are even lower.
So again going one layer deeper how did we start in understanding and implementing this so we really tried to get a snapshot of current technologies so on the left side we worked with a Cadillac super cruise and so that has lane following and hands-free driving capability on highway. We saw about 100 watts of automation load here so again we're thinking 100 watts versus 2000 watts it’s a very different amount of energy required for some of these simpler functionalities and functionalities that we're interested in from an eco-driving approach. Similarly, we partnered with fev on a smart demonstrator vehicle that afforded a highway hands-free as well as some attention-free driving and it also did low speed fully driverless operation. So we ended up seeing about 380 watts for this vehicle and then really understanding again it’s an important spectrum of capability and then being able to match where you need the power and you to do the processing really putting a flag in some of these new capabilities.
I really want to highlight the piece here that I think at the bottom is automation loads are really evolving quickly there's a lot of balance in terms of these numbers so we're trying to balance safety additional functionality whether it’s higher resolution cameras or additional sensors and then we're also looking at processing improvements so machine learning. In general and all of the connected vehicle and automated vehicle stack is dramatically increasing its processing capabilities and energy consumption question so really tracking this is sort of the first piece or first level of understanding about all of these different spectrums. But it’s helpful to have this test data really start thinking about the wider spectrum of operational envelopes moving on to the second piece so a lot of details were talked about this in the Smart Mobility modeling workflow presentation but I want to highlight some of the pieces that the connected automated vehicle color talked about as well so we obviously are looking at single component operation from the ADAS systems we talked or the automation systems that we talked about a second ago then we're looking at individual vehicles.
We're looking at networks of vehicles and traffic flow, so as we get data from different automated vehicles in the field we can then update some of these models and understand how they flow sort of from the vehicle level and the component level. This of course then integrates within the much larger workflow and can be highlighted in Eric's previous presentation on the workflow one example I wanted to talk about here just in terms of the evolving simulation portfolio though is roadrunner which provides a trip level simulation of both powertrain and driving dynamics for energy focused CAV controls development and evaluation so if you think about connectivity and automation. Not only do we have awareness about the vehicle's current and future operating conditions and states we also have the ability to change the vehicle's trajectory itself so now vehicles can interact with each other they can speed up and slow down regarding how they want to interact with other vehicles and they can also be aware of what the future and current operating conditions are. So, there are a lot more levers of optimization and a lot more controls flexibility in terms of what you can actually do to get some of these CAV enabled controls benefits. So that of course requires an evolving workflow, and this is just one of the pieces I wanted to highlight there. I think in terms of the overall workflow that's shown here it’s really interesting to start using real data driven drive cycles and maps not only are they moving toward the location but also dynamically understanding how the vehicles will drive during their actual position in these dry cycles. Similarly we want to understand connectivity from traffic cycles we want to understand looking at different vehicles we want to understand classes and then build upon the powertrain simulation capabilities the DOE is very strong with and then we want to understand levels of connectivity controls and all of these pieces sort of integrated and working together in a coordinated workflow of course then we want to automate this do post-processing and really start getting into analysis of the results. So again this is the evolve portfolio as an example but I think there's a lot of work that the CAV pillar also did for micro simulation as well as some of the regional modeling capabilities understanding how these vehicles work from a data driven perspective and then building that into the modeling workflow
So now we're moving on again sort of these research questions three and four. The different focus areas this is a suite of three examples that sort of show the interesting and difference types of analysis that you can do with the individual vehicle level as well as the system level and then really harnessing connectivity. So, the first example is in collaboration with the Volvo drive need field trial work that was done in Sweden. We looked at ACC so adapters cruise control vehicles operating within their specific region we saw a fuel consumption decrease about five to seven percent for these vehicles at an individual vehicle level. So, it’s a promising technology but if we then look at some micro simulation based at high penetration levels of adaptive cruise control we see negative impacts on traffic. So, lack of communication and delays induced by the adaptive cruise control systems they end up leading to traffic instabilities and congestion. So, we actually ended up in a micro simulation study we see a fuel consumption increase of about 60 percent so that really starts to help identify the dynamics between individual vehicle benefits and system or multiple vehicles working together benefit.
Ultimately we talk about a collaborative cooperative solution to some of these issues, so if we have cooperative adaptive cruise control, so now vehicles can talk together and coordinate and understand the surrounding conditions we really now see a lower energy use. So not only are we combining communication we're also having that automation fused with the communication. So we can look to up to a 20 fuel consumption benefit and that again is looking at communication and refined control so it’s really interesting to start from individual vehicles benefit from just this automated cruise control but as we drop that into a system of vehicles we see dramatically increased fuel consumption but then if we start harnessing the controls and communication capabilities afforded to these vehicles, we can start getting benefits again from the different the different strategies and operational conditions.
Similarly we also have a lot of operational controls about both vehicle and powertrain controls and so one of the major keynote findings of this work was being able to adapt to conditions other vehicles and traffic lights by not only changing the vehicle control, so vehicle trajectory itself but also the power change powertrain control is a very effective strategy and so we see a dramatic up to a 20 increase in individual fuel consumption reduction for some of these technologies. But then when we start understanding we see start to see variability across different driving styles just like David mentioned into the introduction we see differences during thriving conditions. We see differences about the relative impacts of automation and connectivity and so it’s really understanding the full space of these different technologies and their impacts across different facets of vehicle technologies. But also understanding what's possible at the high level so, again maybe peeling one more layer down into this I really want to highlight this was a really comprehensive very large-scale study of real-world implementable controls so we're really talking about a large scale simulation study over 4,000 simulations were actually run so we really started looking into things like powertrain type and technology scenario because the controls and benefits afforded to conventional vehicles electric vehicles hybrid vehicles as well as how the change how the technology changes impact the value and benefit proposition of connectivity automation is really important to understand for future studies.
And also, technology portfolio planning similarly the control types and you can see a little bit of this exploded on the right-side view. So first of all we can look at just speed only optimization so how do we control the vehicles to minimize attractive effort and also the second version is how do we not only control the speed but also the power train control of the vehicle to actually look at minimizing things like total electric consumption for a total fuel consumption for a conventional vehicle. So it’s really understanding the difficulty and the advantages of adding the speed and powertrain control versus just speed only control relatedly we also can talk about different scenarios of infrastructure to vehicle communication as well as different scenarios what happens when you're leading a vehicle or what happens if you're following a vehicle. it’s really interesting in terms of how the benefits of one optimal vehicle may or may not sort of filter through into the rest of the vehicles operating in its nearby surrounding area similarly really understanding how to look at real world driving. I think that's a key piece of information when we talk about connectivity and automation how do we look at the realistic mix of driving conditions but also look and generate realistic driving profiles for these vehicles when they're operating in different locations different areas in different regions of the country that have different traffic flow patterns.
Not only do we see benefits and interesting sensitivities at the vehicle level, I really wanted to highlight again the optimal control case also impacts component usage and efficiency. So, when we start delving into these questions there are a lot of important and interesting dynamics both from the connectivity and automation perspective but also in terms of DOE's overall portfolio of technology. If connectivity and automation change the emphasis or importance of certain technologies, we want to be aware of how that sort of evolves the portfolio in parallel with all these ongoing connectivity and automation benefits. So, on the left side we can see some results from the large-scale case study specifically looking at a system we can see quite a few more gear shifting from the optimized driving and speed powertrain and speed optimization case. We see a lot more shifting partly because we know more about upcoming environments but also because we are enabled to have a little bit more flexibility in all of the awareness that has come into the vehicle. We can really understand how to optimally drive and shift and operate the powertrain relatedly just like we talked about for the gear shifting. We see higher much higher motor efficiencies as well for these baseline versus from the baseline versus optimal cases so we're seeing dramatically improved operational conditions at the component level as well as the overall vehicle level, so it’s really important to really identify. Are we using certain components more or less in a connected and automated future? And how are we pushing the envelope of where we're using those conditions. so if you think about how that impacts the total technology portfolio it’s really interesting to understand not only the high level pieces but also getting that all the way into optimal controls and how those optimal controls impact the different value propositions of the different components.
Now we're moving to a more complicated scenario where we're looking at merging. So, we have optimally controlled vehicles that are now merging into traffic of various penetration. So, this is a really interesting study again highlighting some of the questions that David mentioned earlier. What are the sensitivities to penetration level? What are the sensitivities to traffic conditions and how do these vehicles sort of integrate themselves into the overall flow of traffic? So this is a study that actually looked at a seven mile corridor an i-75 of Tennessee and we saw at twenty percent penetration as connected and automated vehicles we saw about a four percent reduction in overall fuel consumption at a hundred percent penetration we saw about a seven percent reduction in overall fuel consumption. So, there are a lot of interesting sensitivities and insights from the study one of which is even at low penetrations we still see appreciable benefits for automation and connectivity. This is real really promising in the sense that we see benefits all the way through an adoption trajectory. We don't just see the benefits at 100 penetration similarly again seven percent we look at overall benefits for CAV penetration. Again, delving one layer a little deeply in here we started with an optimal coordination framework that looks to minimize the overall attractive effort for the area under study.
Of course, then in a traditional optimization approach we're looking at vehicle dynamics boundary conditions and safety constraints in terms of vehicles not encroaching and not running into each other. We then build that into the control zone and the merging zone and on the bottom. I really wanted to highlight the progression, so this is really the result of a three-year progression of work we started with simplified on-ramp merging we then started to build that into more realistic on-ramp merging. You can see an image of the ACM facility here in terms of what some realistic freeway merge scenarios look like and then we pull that all the way to a the i-75 corridor. So not only do we have a realistic merge scenario we have real traffic conditions in a real corridor condition. So, it’s you know starting at the fundamentals and understanding how to develop these coordination frameworks and then expanding that out to assess the benefit and realistic operating condition.
So again, just to quickly summarize there are a lot of really exciting results. A lot of really interesting insights at the very high level and taking that all the way to the very low technical level in terms of detailed insights about different technologies and different control strategies. To quickly summarize again from the experimental side we see for example a 7 to 13 benefit for truck platooning and that was done by a very comprehensive set of work and experimental testing that was done with real vehicles real prototype vehicles very intensive instrumentation. Similarly, we see accessory loads about 400 watts or even lower from existing vehicles. So a prototype vehicle as well as a in production vehicles so lots of interesting observations about the functionality and loads that those vehicles are seeing then moving over onto the controls and automation side we see about a 15 to 20 percent fuel savings benefit from cooperative adaptive cruise control. So, remember sort of the suite of going through individual vehicle benefits may not necessarily translate to overall benefit, but cooperation through automation and connective connectivity is really important for some of these benefits. Relatedly if we start combining vehicle and powertrain controls, we see up to a 20 benefit in fuel consumption reduction. So really understanding all of the levers and all of the capabilities that connected and automated vehicles entail and then of course coordinated merging that we just talked about we're seeing a four to seven percent benefit in fuel savings. Of course, this changes depending on scenario as well as penetration but really enabling overall congestion and consumption reduction through connectivity and intelligent use of connectivity.
With that again I really want to highlight this was really just a very quick brush through a lot of the thinking and a lot of the approaches that came through from the vehicle. The CAV focus area principal investigators contributors I’m really just sort of explaining some of the work and thinking that went into it. It was a very diverse and broad-ranging set of researcher’s capabilities in labs that was really brought together for really exciting and really ground breaking work in terms of connectivity and automation. So with that I will hand it back over to David and we'll talk maybe a little bit more for questions and open up for questions and discussion and I really want to highlight this is really a very a very high level overview of some of the thinking and some of the high level conclusions. So please refer to the capstone reports and the web the website for these seminars to really get a much deeper understanding of some of the issue’s questions and really exciting insights that came from this work. So, thanks to everyone for their understanding and a listening and I open up to some questions.
David Anderson:
All right, thank-you very much, Eric; great summary of a lot of work that if we dove into all of it would take a day or many days to fully cover. As we, as we transition to our Q&A period again I’ll remind you if you have questions please enter them into the Q&A box on your Webex screen and we will get to as many as we can. I will also mention that you know as Eric began his summary of some of the results from the CAVS pillar of research and Smart Mobility, he began by summarizing some of our platooning results focused on heavy-duty trucks. So, I I’ll remind you all that VTO currently has an RFI or a request for information out on the street. An RFI focused on soliciting feedback on our medium and heavy-duty truck research and development activities that RFI is number 2372. If you're familiar with how we number our RFIs and our FOAs and our notices of intent that's the last four digits. So DE-FOA-0002372 if you want a URL to pull up that request for information on medium and heavy duty trucking please ask for it in the Q&A box and we can provide it to you.
So, with that we'll move on to questions we have a question. Eric, I’ll pose it to you, but I’ll give a little preamble to it. The question is how do you forget protect against cyber-attack? Obviously when you start talking about connected and automated vehicles you know connectivity represents a, you-know an attack surface, a threat surface and I tend to think that automation sort of amplifies the impacts of that threat. If it’s compromised so very important question however it was out of scope for Smart Mobility 1.0. Eric, do you have additional response to that question?
Eric Rask:
Yeah, so, I will maybe just build on your out of scope question. So, it was it was definitely something that was out of scope and smart 1.0 and I would highlight there our focus was really on understanding what's possible from connected connectivity and automation. So, starting at that high level view and then maybe thinking about ways that we need to protect or we need to change sort of the implementable and optimal ways to put those systems in access. So I would just highlight there's a lot of really exciting cyber security capabilities and work going on across the daily lab systems and so I think I expect to see more integration of some of those slot pieces and thinking in the future from the elite VTO efforts. You know the way I like to think about it is you sort of have a map of optimal benefit and then maybe something like cyber insecurity or degree of handover to an unknown entity and you can really start prioritizing and balancing places where there is a very large cyber security risk for minimal benefit, or a very large benefit for moderate cybersecurity risk. So really starting to prioritize and bin those from the benefit challenge perspective and then build from there.
David Anderson:
All right, thanks, Eric. We have another question. This is a great question so “for the analysis of the ACC vehicles in the multi-vehicle scenario, were the control algorithms from actual OEM systems employed?” And that question goes on to ask, “do different control strategies impact the fuel consumption and how much variation is there among different systems”. And so again the question is kind of you know did we use OEM representative ACC controls in doing the study of fuel impacts of adaptive cruise control systems?
Eric Rask:
Yeah, absolutely, so this study was in partnership with the Volvo drive me pilot. So they were adapters cruise control controls that were developed by Volvo and were running in a pilot level so they certainly had a lot of the detail and information that a normal OEM would use for something that was piloted and field trial. That said I agree with the point that really understanding the variation and maybe preferences manufactured to manufacture is important. And so I know that there's some ongoing and future work funding provided from the Department of Energy that is looking into some of these questions is really how do we get the breadth of ACC. Because when we say ACC that doesn't necessarily just mean a single vehicle it may mean a whole range of vehicle behaviors. So I would just highlight that it was in partnership with an OEM but I think we're cognizant that there's a wide variety of behaviors and capabilities from even just the ACC system so I think that's definitely a topic for further study and integration and then validation of course into the modeling portfolio.
David Anderson:
So, speaking of validation, yeah, the, there's a question about the control system. Some of the other control systems that we tested primarily in simulation these are lab developed CAV algorithms cap control algorithms and so we have a lot of simulation results. But how are we validating those simulation results?
Right, so I would highlight some of the ongoing work there's a lot of supplemental efforts with the Department of Energy with a collaboration with the American center for mobility and the labs themselves who develop these different CAV controls. So, we really are pushing to in the field at ACM validate and develop these controls to really sort of circle back all the way to the research focus one of how do we develop these prototypes. How do we get data and then how does that filter in? So very much a circular experience in terms of developing the controls and then circling back to prototype development so really exciting work going on with the labs in collaboration with American center for mobility to really have these vehicles operating in the real world and in real experimental conditions and scenarios for validation data.
David Anderson:
Yeah, thanks for that, I’ll add to that you know what we've what we've learned as we start transitioning from a virtual world. The simulation world to implementing these controls in vehicles on test tracks like at ACM you learn a lot when you go from simulation to implementation and so we're focusing resources now on kind of a whole simulation through lab to track pipeline. Focusing on anything in the loop so that we can start to implement these controls in real time in real vehicles in the lab and get them working in that environment. A much safer less costly environment before then transitioning to the track test that we're doing with in partnership with the American center for mobility. So yeah, a lot of work to be done. There is a question here it is: “in automated vehicles how much stress is reduced on passengers inside the vehicle as compared to driver driven vehicles?” I’ll give a little preamble to your answer here, Eric, as well, so I well, I would assert that that's not something we looked at in the CAVS pillar itself. Eric you might have more on that we did have another pillar of research focused on mobility decision science.
Essentially the human role in the mobility system and how humans make decisions about many things like where to live. How to get to work. What technologies to adopt. What car to buy or not to buy. What mode to take and so some of that human behavioral thinking sort of would impact you know, if humans would become stressed inside an automated vehicle they would tend to not purchase automated vehicles or use automated vehicle services. Right if the stress situation is lower than that would lead to higher adoption and so I will use this as a an opportunity to plug our webinar in a couple of weeks focused on moving people in the Smart Mobility system. But Eric I’ll turn it over to you to see if you have any additional comments on that question. Yeah David I would really just echo your points a lot of the initial work is done through the MDS pillar in terms of understanding the value of travel time and as enabled through automation. But just as a you know, future research topic and future research question I personally as a person who gets car sick very frequently, I’m always interested in how automated driving differs from you know manual driving in terms of how people can feel carsick or what type of behaviors lead to sort of nausea, versus like OK driving and doe that change and then how does that interact with the efficiency strategies that are being done here.
So definitely out of scope of the preliminary look but I think it’s definitely a really interesting topic in terms of how people feel and interact in these vehicles. Just anecdotally I know some people when we had experimental vehicles that were making sure that you were paying attention to the road they actually felt more stressed driving there because they seemed like they were always being watched on some level in terms of watching the road and sort of being made more aware that they needed to be attentiveness. Just from the human perspective so I think it’s a really interesting question but very much a future research topic and figuring out where DOEs role fits there. Yeah thanks I’ll add to that I mean that sort of reinforces our whole need to move from purely simulation to testing and validation. You know we can we can create very efficient cab control algorithms that work great in simulation but there are drive quality and ride quality and issues just that you just pointed out that you might not necessarily detect in simulation but become evident when you start operating the vehicle on the test drive. And so, some of the initial simulated results might even over exaggerate or overestimate the benefits that you can actually get in the real world.
All right, there is a question, so the question is in terms of intellectual property about the information that systems provide are there any legal or ethics concerns? So I will take that question to mean rather than intellectual property I interpret that to me are there are there you know data security issues by the information that might be being passed from one vehicle to another.
In a V to V or between vehicles and infrastructure and in V to I do you have a comment on that, Eric?
Eric Rask:
Yeah, I think they're very much our issues. I think it’s you know again maybe out of scope with smart 1.0 but I think that's an interesting dynamic and I would almost add that as the second leg of the cyber security question. So not only is there cyber security but then there's privacy and then there's let's call something like OEM or control strategy privacy in terms of very valuable information and IP actually built into these systems. But those systems are then operating and communicating in a much more open world so I think it’s a very interesting question but certainly something that we didn't do in smart 1.0 because we're really focused on really assessing sort of the opportunity space and sort of the benefits and sensitivities very much an interesting dynamic.
David Anderson:
There's a question related to the electrical power loads you evaluated. The you know the additional electrical loads that are required for different levels of automated systems. Yes first part of that question is “what is responsible for those increased electrical loads? Is it the is it the sensing part is it the computational part what where do those increased loads come from? And then secondly is there additional research needed in order to reduce those electrical loads?” Yeah that's a great question! So, I would say actually the answer varies a little bit depending on the functionality.
Eric Rask:
So, on something like a Super crew system, we see a balance between additional loads associated with the sensors providing the awareness and that's actually really driving a lot of those loads. But then we see of course some additional loads from the processing side of things, but we don't see as much now for moving to these higher levels of automation. Specifically like the SED demonstrator or a fully automated vehicle we see the balance of where that excess energy is coming from really moving toward the processing. So now we're really seeing a lot of the energy loads being driven by the processing associated with object detection awareness, path planning all of those different issues. So that dynamic shifts to be a lot more driven by the processing system. Part two of that is again it’s a really rapidly evolving space. If we think about some of the capabilities in the Tesla autopilot system versus some of the other automation systems and some of the claimed load levels that are there we see a really wide range of power consumption across those systems. So, it’s very much a research topic. I would say not only in the vehicle space but of course low power machine learning and AI is a topic that spans supercomputing work done at the DOE, it spans consumer goods like phones and so you know there's a lot of tailwinds across the research industry. But it’s very much a question because I think the imperative is we want to make sure that we can seize the benefits due to connectivity and automation but we don't want to have those benefits washed away by the excess computational loads that are required for some of these more advanced features.
David Anderson:
Great, yeah, another question that that makes it such a complicated space we have a question it’s: “the current focus is mostly on fuel saving and energy saving have you considered other factors such as emission reduction or environmental impacts from the application of CAVS?”
Eric Rask:
Yeah, I would I so the primary focus was on congestion and energy reduction but certainly I think there's an expanding view of what mobility means in terms of how do we really assess that and we talked a little bit about that with the MEP score earlier. But the focus initially was on energy and congestion but certainly a lot of the simulation capabilities as well as the experimental data also does have some opportunities to look at emissions and overall environmental benefits. So, it’s just starting and so not only maybe the emissions, but I think also understanding the life cycle analysis impacts of some of these components. So, I would say it’s very much an expanding and evolving research topic within the Department of Energy. That said DOE has a lot of expertise in understanding emissions in understanding life cycle analysis. So, I think it’s really building into that. I don't know if David has another comment to add there but I think it was initially focusing on energy and congestion but then really expanding that that portfolio. Yeah as you were as you were answering there I was I was thinking I might amplify your answer a little bit when you're done so exactly what Eric said, our initial focus was on energy. We are the Department of Energy after all. But as we've expanded our purview our scope to look at mobility more broadly, as Eric pointed out we have this metric called mobility energy productivity and so that looks at energy but it also looks at the cost affordability and it also looks at the time required to access opportunities.
So, you know ultimately our goal is to reduce the cost and the energy in the time associated with accessing the opportunities. The work, the play, the school the groceries… things like that define a high quality of life as we you know. As I mentioned in the introduction this is a large multi-disciplinary complex sector. There are multiple agencies working on it including DOT and we're engaging with our colleagues over at USDOT and EPA, obviously has a strong mission focused on emission reduction and so you know we discussed with them how some of our modeling tools which are being used to quantify energy can also be used to quantify impacts in terms of emissions. You know criteria pollutants there are yeah there are many benefits that accrue from improvements in the transportation and mobility system beyond energy and while we started with energy we are you're certainly cognizant in trying to do what we can to support missions that have other metrics of concern. So, thanks for that question we are getting very close to running out of time one more question.
You talked a little bit about the road runner tool that's being used to simulate not just individual vehicles but vehicles interacting with one another. In past webinars we've talked about some of the tools as well and there's always a question about how are these tools being made available to other folks so can you talk a little bit about the deployment of the road runner software that you mentioned. Yeah so specifically our Argonne National Laboratory is developing the road runner software and it’s currently available for partners for evaluation feedback and understanding of you know really getting it into a useful system and tool ultimately it’s slated for release through the amber tool framework. Which incorporates a whole bunch of other argon simulation capabilities next year. So I think it’s available again right now for partners and people working on research projects and then it’s going to be rolled out through the amber framework in the next year so I would really recommend looking and working through the /eere/vehicles/energy-efficient-mobility-systems website and trying to work through David for any requests or any thoughts about more information there. Yeah feel free to contact us through that email address that's on your screen for more information.
David Anderson:
Maybe time for one more question, the question again based on trucking. So the question here is at least one truck manufacturer has backed away from platooning or what the question says for practical reasons, for practical reasons, can you comment on what might be done to make platooning more acceptable and attractive?
Eric Rask:
Sure, I think of course it’s always a balance of benefits and business strategy and commercial strategy you know there's a lot of different facets to understanding that. I think some of the controls developed on the argon or excuse me on the Smart Mobility side really looked at how do we make the vehicles more robust to cut in and more robust to sort of these real world issues. So I think there's a lot of open research questions about making the systems more robust and then I think there is truly a question to understand the commercialization and value chain of where platooning and some of these strategies make sense. Specifically if you're talking about how do you coordinate vehicles that are owned and operated by different entities, how do you coordinate who something as simple as who leads the platoon and who follows the platoon how do those systems coordinate. So I think part of it is robustness to you know real world operating strategies and conditions and then part of it is really understanding and collaborating industry government academia how do we understand this value chain and understand what all the stakeholders and players are.
David Anderson:
Yeah, thanks for that, Eric. I yeah, I agree, I think it’s you know our role is to provide information data on the effectiveness, the opportunities associated with platooning. I think it’s up to the industry to make their decisions regarding if it’s a technology pathway they want to pursue. So, with that we're at four o'clock eastern we'll go ahead and close out. I encourage you all to visit the link that is shown on your screen here where you can download the six capstone reports that were published just a few couple months ago. Five reports one on each pillar of the Smart Mobility research effort and then one report on the modeling workflow development and results again you can do that at the website you see here you can also register to attend our remaining webinars which you see on the screen here at that same URL. Our next webinar will be in two weeks, November 19, where we will focus on moving people a Smart Mobility system. Two weeks after that we'll focus on moving goods and freight in the Smart Mobility system and then finally, we'll wrap it up on December 17 by looking at ep chart -- excuse me, EV charging infrastructure needs in a future Smart Mobility system.
Thank-you all for attending the webinar today. We hope you found it informative stay safe and we look forward to you joining us again in two weeks thank-you, everyone!