Category Archives: ACV Conclusions

Autonomous and Connected Vehicles Report: Key Findings and Gap Scoring Methodology

The following are priority recommendations about R&D gap targets identified during the FY16 and FY17 research period.   These recommendations are based solely on Synthesis’ conclusions and do not reflect the viewpoints of the VTO or any particular source.  Anonymity of primary sources is maintained unless Synthesis receives specific approval to share this information by individual sources.

Results from FY16 Work

Synthesis ranked ten top-level R&D topics raised by primary sources (based on open-ended, in-depth interviews), intended to represent a broad range of interests and technical fields, based on four variables using a basic scoring system.  The four variables used for ranking are a) expected energy efficiency improvement; b) amount of scientific evidence provided by sources; c) relevance to fast-growing US industry sectors; and d) a Google Scholar activity index.

The top-ten R&D topics identified from Synthesis’ contact with 100s of sources are:

  1. Embedded software
  2. Variable compression ratio
  3. Motors
  4. Power electronics
  5. Car sharing
  6. Autonomous vehicles
  7. Vehicle embedded software
  8. Solid-state batteries
  9. Carbon fiber
  10. Advanced cylinder deactivation

In terms of highest-priority recommendations, Synthesis drilled down to identify the following three areas as both directly relevant to future automotive technology capabilities and among export categories in which the US shows unique, industrial strength:

  • #1: Automotive semiconductor manufacturing and manufacturing equipment.  This includes semiconductors designed for discrete and integrated circuits (ICs) for power management, electric traction drive power electronics (inverters and converters), as well as signals processing for advanced driver assistance systems (ADAS), and infotainment systems.

  • #2: Embedded software. This covers software written to execute a particular function in a particular hardware implementation, such as field-programmable gate arrays (FPGAs), to include functional capabilities that are increasingly important for “smart” vehicles and automotive electronics, such as:
    • Machine vision;
    • Image analysis;
    • Signals processing for self-driving cars;
    • Safety management;
    • Wireless updating;
    • Artificial intelligence;
    • Automotive cloud engineering; and
    • Software to drive a wide range of field-programmable gate arrays (FPGAs) that can be altered based on the application.

  • #3: Integration of power management and wireless baseband in application processors.  This covers system-on-chip (SoC) design, in which an IC integrates multiple components of a computer or other electronic system into a single chip.  It may contain digital, analog, mixed-signal, and often radio-frequency functions – all on a single chip substrate.  SoCs are very common in the mobile electronics market because of their low power consumption.  They are becoming the first choice for future embedded system developments – which are of increasing relevance in automotive applications – due to the increasing demand for higher performance, reliability and power density in autonomous applications.

It is notable that hybrid and electric vehicle manufacturing are driving vehicles toward computerization.  Microcontrollers, sensors and analog devices have led the growth of automotive semiconductors.  McKinsey analysts have noted that three areas will drive the next wave of growth: 1) further electrification of drivetrain; 2) ‘consumerization’ of auto electronics, and 3) vehicle intelligence (including active safety innovations and connectivity-enhanced driving).  The electrification of the drivetrain due to hybrid or fully electric vehicles may lead to the “largest expansion over the next ten years,” according to these analysts.  The drivetrain accounts for 30% of all semiconductor content in an automobile, and “developing a less expensive alternative to IGBTs would be one way to win the market share in this area.” [1]

Automotive power management is the third largest market for semiconductors, comprising about 8% of the total market in 2015, with a 10.8% growth rate compared to 5.5% for the total IC market.  In 2014, the automotive market for ICs totaled about $21.7 billion.  Amongst automotive power management applications, demand is strongest in advanced driver assistance systems (ADAS), which include Lane Departure Warning (LDW), Forward Collision Warning (FCW), Automotive Emergency Braking (AEB); and infotainment systems.

From a market perspective, the worldwide automotive semiconductors industry is more than a $24 billion business and has experienced one of the fastest growth rates of any large segment in the $300 billion worldwide chip market – averaging 8 percent annually between 2002 and 2012. The top-ranked companies involved in semiconductor equipment manufacturing are provided in the table below.  US headquartered firms in this sector accounted for 39.5% of the global market in 2015.  This demonstrates an apparent competitive strength of US firms, from which Synthesis is working to identify new R&D topics relevant to the VTO mission, with transition opportunities.

Top 10 Worldwide Semiconductor Manufacturing Equipment Vendors, By Revenue ($ Billions)

In the automotive semiconductor sector there are certain technical gaps that – if filled by US-based engineering R&D advances – present a concrete opportunity to expand US-based advantages in design and development of critical engineering systems and products for the automotive industry and beyond.  US-headquartered firms are among the top-ranked companies in the automotive semiconductor sector and accounted for 39.5% of the global market in 2015.  For this reason, research work in core R&D in this sector, including research on next-generation energy efficiency targets for automotive applications, has inherently strong transition opportunities.

Within the semiconductor space, Synthesis assessed in FY16 that the following technical and engineering fields are notable for further discussion regarding R&D gaps.

  • Functional chip-designs for automotive manufacturability.  For example, Jen-Hsun Huang, [then] chief executive of graphical chip maker Nvidia Corp., said [in 2016] that some of its’ processors were being strained as Tesla has increased its cars’ capabilities.
  • Instruments, methods and designs that ensure chip-design complexity keeps pace with, rather than outpaces, manufacturing productivity.  Put another way, the present challenge is that chip-design advances are outpacing manufacturability advances, and thus, for example, lithography presently appears to lag the growing need for complex designs executed on smaller surface areas at lower costs (see Synthesis’ FY16 extreme ultraviolet (EUV) case study).   One increasingly can design chips that are needed, but cannot be produced at a competitive cost – and this presents key research, development, engineering and testing gaps.
  • There is a continuous need for developmental work in multicore system-on-a-chip (SOC) architectures, to enable faster and “smarter,” increasingly functional, smaller, more power dense electronics.

Results from FY17 Work

At the beginning of FY17, VTO tasked Synthesis to further frame the opportunity for future R&D targets based on findings from FY16, and to focus in particular on the areas of autonomous drive, LiDAR (light detection and ranging) sensors and connected vehicles. 

Synthesis identified 37 R&D gaps in FY17, from the integrated primary and secondary source analysis, from February through August 2017.  Synthesis assessed numerous trends and technologies related to autonomous drive, sensors and connected vehicles.  Findings are derived from a structured, data-driven process intended to produce findings on plausible, high-value R&D targets for sustainable, US-based jobs.

Each R&D gap can be traced back to primary source interviews or secondary source documents.  A gap is based on specific needs stated by key sources, “in ongoing commercial R&D and product development activities,” which pertain to autonomous and connected vehicles that is:

  • Hardware-focused;
  • US-based, or have the potential to be US-based;
  • Could reach commercial vehicle markets in 5-10 years; and
  • Has the capability to reduce costs, ideally by a significant (>50%) amount.

The 37 R&D gaps are categorized and scored, with trace-back capability for each gap and score to Synthesis’ internal data sets.  The quantitative analysis of the gap topics is offered as an initial viewpoint. Synthesis welcomes discussion on additional perspectives to assess the nature of the gap intelligence obtained.  The following summarizes the findings, assesses the distribution of gaps across categories, and provides an analysis of the gaps based on quantitative scores.

Drill-Down on 37 R&D Gaps Identified in FY17
The 37 R&D gaps identified in FY17 are grouped into three categories developed by Synthesis to provide guidance on trends in findings, as follows:

  • Vehicle Sensors and Intelligence Materials: Hardware for low-cost, high-performance, energy dense on-vehicle data storage, processing and communication on the vehicle; includes LiDAR.
  • Vehicle-to-Vehicle (V2X) Communications and Intelligence Networking: Hardware for low-cost, high-performance, energy dense, secure and reliable communications V2X, including sensors and sensor fusion.
  • Other R&D Collaboration Opportunities: Other opportunities to address R&D gaps in the autonomous vehicle and V2X

The three bins illustrate the range of R&D topics that Synthesis identified as gaps during this research.  These categories are not mutually exclusive and are certainly not the only categories that could be developed from the gaps.  Follow-on analyses that assess the relationships within and between smaller groups of gaps for the purpose of identifying (more efficient, multi-impact R&D topics in more detail can be done.

As just one example of such a drill-down on gap categories, and focusing only at a high level based on numerous conversations with sources, there is an apparent relationship between the individual gap categories used in this report and the level of interdisciplinary RDT&E (research, development, testing and evaluation) work that is needed to respond to each gap.   This view is reflected in the Figure 2 below.  As research moves away from addressing gaps in (e.g.) individual sensors or sensor materials, and toward (e.g.) V2X communications systems and complex systems, the work required – speaking in general, at a high-level – is increasingly of a multi-disciplinary nature, moving beyond hardware alone.  More research on the validity and implications of this finding is recommended.

Gap Scoring Methodology

A scoring system was developed to analyze the relevance of the 37 gaps identified in FY17 and to introduce initial rankings to address the work-plan objectives.  Additional scoring techniques are feasible and available for discussion (e.g., scores that apply to clusters of gaps or specific technology attributes of gaps).  To start the discussion, the following three scores are employed:

Score #1: Work-plan relevance:
For each attribute, which included (1) hardware-focus, (2) more than five-year relevance, and (3) private sector R&D investment gaps, the following scores were applied:
0 points assessed if the attribute does not apply;
5 points assessed if the attribute partially applies; and
10 points assessed if the attribute fully applies.
The maximum score on work-plan relevance is 30 points.

Score #2: Number of sources in agreement:
1 point for each company or individual expert source in agreement;
25 points if the gap is developed from an AVS 2017 consensus finding, (approximates the 25 experts, conservatively speaking, holding the consensus view).
The minimum score is one and the maximum is in increments above 25.

Score #3: Sum total:
Summation of Scores #1 and #2 above is used as the integrated Synthesis final score for final rankings.  (Again, this is just one way of ranking the gaps and additional scoring approaches are available for discussion.)

Levels of Hardware or Software, and Interdisciplinary RDT&E (Research, Development, Testing and Evaluation) Work, by Category

Distribution of Gaps Identified in FY17, by Category

The 37 gaps were analyzed and placed in one of three categories.  Of course, different categories could be developed depending on how detailed one is in assessing each gap, or group of gaps.

30%: Vehicle Sensors and Intelligence Materials
30%: V2X Communications and Intelligence Networking
40%: Other R&D Collaboration Opportunities

Autonomous and Connected Vehicles Report (2016-2018)


Gap Analysis

Final Recommendations and Conclusions

Autonomous and Connected Vehicles Report: Final Recommendations and Conclusions

Major Recommendations

The following summarizes recommendations from Synthesis’ FY17 work.  These recommendations focus on autonomous vehicles, connected vehicles (vehicle-to-anywhere (V2X)), and LiDAR sensors.  They provide the “consensus” viewpoints, based on Synthesis’ ranking of information collected from 100s of primary sources and 1,000s of secondary sources evaluated during FY17.

  1. There is a clear opportunity and need to address the challenge of integrating software and hardware for future mobility applications, particularly mobility use-case-driven R&D gaps in software-enabled, V2X systems are expected to grow significantly.

  2. A taxonomy or roadmap of R&D gaps in autonomous and connected vehicles is needed to depict the attributes and categories of technical gaps.  In this regard, there is an opportunity to lead R&D data collectors to collect, categorize, and quantify the priority or relevance of specific types of R&D gaps– especially in the fast-changing and growing LiDAR, sensors, autonomous and connected vehicle research spaces.

  3. Support for engagement in hardware-oriented sensor R&D, including in:
    a. Substrate-level (e.g., increase wafer size) R&D
    b. Embedded electro-optics design and engineering (e.g. improve imaging resolution, resilience, accuracy)
    c. Glass materials development (e.g., address reduction of REs, and add improved optics and machining properties)
    d. Application-specific electro-optical engineering (e.g. learning-by-doing manufacturing and electrical engineering for human transport applications, in terms of cost, quality and performance)

  4. Two gaps in the V2X and Intelligence Networking category are consensus gaps and deserve VTO’s immediate attention:
    a. R&D on Low-Cost Geo-Localization with 2-3 cm accuracy: Opportunity for design and engineering of low-cost sensors and sensor fusion systems that enable required accuracy for reliable autonomous vehicle applications.
    b. V2X Software-Hardware Integration: Opportunity for guidance, independent test, validation and understanding for software-hardware integration for highly specific AV use cases.

  5. The following top five consensus gaps in the Other R&D Collaboration gap category are recommended for VTO’s immediate attention:

    a. Fundamental Competition at Core Technology Level (e.g., Robotics, Machine Vision, AV Sensor Requirements):

    The need is for more clarity, commitment, and investment regarding the core figures of merit, the baseline current state and the targets for future R&D – all to accelerate potential breakthroughs in these fields.

    b. Scalability of Autonomous Vehicle Engineering (Hardware, Software, Multi-Context, Global Implementations)

    The need is to investigate AV engineering approaches across a spectrum of operational contexts (e.g., city, urban, people-transport, things-transport, logistics, on-ground, marine, air, etc.) and to address where scalable solutions are being executed, are feasible, and to identify and share best-practices.

    c. Situational Awareness: Data Repository on Technology Used (AVs, Smart City and Smart Grid)

    The need is for a verifiable, publicly accessible data source that provides multiple stakeholders (e.g. academics, researchers, investors, inventors and state-local-federal partners) clarity about who is doing what where, in order to catalyze facts-based decision-making in this important and growing field.

    d. Systematic Assessment of AV Technology Gaps

    The need, as discussed above, is for a systematic, independently derived view on the technology gaps in autonomous and connected vehicle applications. A technology roadmap is recommended based on the high-score for this gap.

    e. Need for Opto-Electronic Engineers (including for LiDAR, Optics, Sensor Development Engineering)

    The probability that AV technologies, like LiDAR, will be available at significantly lower costs for widespread application depends significantly (indeed, this is a first and foremost driver – based on numerous sources) on the availability of a skilled workforce to ramp up production, reduce costs and maintain quality of end products.

  6. The availability of skilled AV-domain manufacturing engineers is viewed as a key constraint to growth in capacity (esp. if the growth is rapid) of autonomous vehicle technologies (including LiDAR) in the US.

    Disruptive Technology Recommendations

    The extensive data collection and analysis in FY17 suggests that there are disruptive technology R&D opportunities to consider in the fields of LiDAR-related sensors, connected vehicles, and V2X.  The data also suggests that there may be unique opportunities for US-based job creation by focusing on such leap-ahead innovations.

    “Disruptive” is defined as technologies which present opportunities to address significant cost- or cost and performance gap closing needs by one or more orders of magnitude, in which current technology is too costly to scale to address the radical performance increases and cost reductions that are needed. 

    Each of the following recommendations regarding disruptive technology reflects Synthesis’ independent assessment of both primary and secondary source research completed under this work-plan.

  7. New I/O control architectures can address multiple on-vehicle intelligence processing functions, at human-transport quality, in fraction of time and fraction of cost.  For example, fusion of all domain control units (DCUs) in one centralized vehicle computer OR new validated, tested architectures that permits “full sense-and-compute” at the edges. New I/O control architectures would present:

    a. A solution that removes the requirement to send data off of vehicle, including for map location, navigation or processing.
    b. A solution that is lighter, faster, more secure data generation, collection, mining and processing on vehicle.
    c. solution that can scale in terms of information processing to match the “every car a map-maker and every car a map-user” paradigm.

  8. Power by Ethernet:” This covers the need to wirelessly charge AVs in many contexts, while mobile or stationary.  An alternative to costly production, install, maintain and package of power through (heavy) motors, batteries and wiring harnesses.

  9. Accelerated engineering of “hard-coded hardware accelerators.” This includes – for example – hard-coded devices that can enable fleet-wide AV software upgrades and that enable such software upgrades to be:
    a. Fail-safe
    b. Valid
    c. Cyber secure

Final Conclusions

The combination of autonomous systems and Internet of Things (IoT) demands new, in-depth understanding of future engineering R&D requirements at every level of the systems engineering process.  Information is (becoming) the new energy.

New partnerships among OEMs, Tier 1-3s, software developers, cyber-security experts, research universities and federal R&D labs are needed to catalyze R&D work in hardware, software and systems-engineering fields.  Such new partnerships are needed to guide the software-based and interdisciplinary work that needs to be done to advance autonomous and connected vehicles.  Synthesis has explored the industry data in this report on a few topics, and has identified several fast-growing fields that will frame the nature of this new R&D reality.

In brief, future R&D will enable systems and components in vehicles to communicate and compute with networks from the component up through the transportation grid, through to a global-level grid.  This is why emerging R&D gaps are numerous, and not only hardware-focused.  From a hardware perspective, autonomous and connected vehicles continuously seek smaller, more functional, more power dense and lower cost designs of every component and sensor.  From an information perspective, rigorous processes for collecting, maintaining and analyzing information about the “information gains” of future R&D is needed.  From a US job and manufacturing growth perspective, this report finds that more skill in engineering-to-manufacturing capabilities is needed.  This simply means more learning-by-doing. 

The fast-emerging, estimated multi-trillion dollar market for autonomous systems is directly connected to (as both a driver and beneficiary of) the Internet of Things.[1]  Semiconductor manufacturers play a key role in this opportunity.

Autonomous and Connected Vehicles Report (2016-2018)


Gap Analysis

Final Recommendations and Conclusions