TU-Automotive Detroit 2016: Perspectives on autonomous vehicle development
Range of companies and entities provide thoughts and technologies about autonomous
TU-Automotive Detroit 2016 exhibition floor
The TU-Automotive Detroit 2016 exhibition was held from June 8 through June 9, 2016, in Novi, Michigan, U.S. Over 3,000 attendees were at the conference, which featured a variety of presentations focusing on autonomous vehicles, mobility, and connectivity. The theme of the 2016 conference was, “Collaborate to put Auto in the IoT (Internet of Things) Driving Seat”. As the automobile is one of the few items in the IoT that can physically move through the world, it can form tangible connections with other items. This capability gives the automotive industry an opportunity to be a key driver in the IoT.
This report will focus on presentations given at TU-Automotive Detroit 2016 related to autonomous vehicles and technology. Information in this report includes Kia and Nissan’s general philosophies and activities towards autonomous vehicle development, technologies that benefit autonomous driving, and perspectives from government entities and companies without an automotive background in regards to autonomous development.
This report is the first of three reports which focuses on the discussion sessions presented at TU-Automotive Detroit 2016. Reports focused on mobility and connectivity will be released at a later date.
OEMs outline autonomous philosophies and plans
Kia summarizes keys behind autonomous vehicle development
|Session: Forget Game Changer: This is a World Changer|
|Kia Motors America||Henry Bzeih||Chief Technology Strategist/Head of Connected Vehicles|
Visualization of key drivers of automated technologies
In the session titled, “Forget Game Changer: This is a World Changer”, Henry Bzeih provided insight to Kia’s perspective on the development of autonomous vehicles. Early in the presentation, Henry Bzeih described four key factors driving the development of automated technologies:
- Societal trends, as related to vehicle accessibility versus vehicle ownership, which can be seen in the growth of services such as Uber, Zipcar, and Lyft.
- The millennial generation is the largest generation in the U.S., and represents a significant portion of the nation’s customers. The generation has its unique tendencies and characteristics, which OEMs need to accommodate to gain them as customers.
- The safety goal of ultimately achieving zero fatalities on roads.
- The development of the Internet of Things.
Understanding these factors provides a clearer view on the evolution and vision of autonomous vehicle development.
In describing autonomous vehicle technologies, Henry Bzeih organized the technologies into four different categories: sensors and recognition, decision and control, HMI systems, and V2V/V2I communications. For a car to be fully autonomous, it requires technologies from all four categories. These technologies allow the vehicle to follow a five-step process to achieve autonomous driving:
Advanced driver assist systems also utilize these steps in their performance, though to a less robust degree than required for autonomous driving. Examples of automated driving features currently included in Kia models include interval autonomous driving, emergency stop systems, traffic jam assist, and narrow path assist.
Henry Bzeih also discussed Kia’s efforts in autonomous vehicle testing and its future plans in autonomous vehicle development. Autonomous vehicles are subjected to harsh environment testing at Kia’s facilities in California and in Korea. Due to recent legislation, the company also performs on-road testing of autonomous vehicles in Nevada. As seen in the automated roadmap given below, Kia was able to demonstrate conditional and high automation levels at CES 2016 in Las Vegas, and hopes to have a fully automated vehicle in the future.
Kia automation technology roadmap
Nissan combines company ideals with mobility development
|Session: Nissan’s Future: Moving Electrification, Connected, and Autonomous Cars to Intelligent Mobility|
|Nissan||John Schnoes||Program Director – Vehicle Information Technology & Autonomous Drive|
Common customer expectation themes based on feedback
Nissan’s presentation focused on how it was incorporating company philosophies of delivering innovation and excitement and a customer-centric approach into its intelligent mobility development. John Schnoes mentioned that the idea of focusing on the customer isn’t only tied into providing features for customers, but also understanding how to convey the feeling of quality to the customer. With the improvements in technology, customers have new expectations of what a quality vehicle can provide. For example, based on feedback Nissan received from its customers, people are interested in seamless connectivity in transitioning platforms from a smartphone or tablet to their car or vice-versa. Customers also want peace of mind from their vehicle, whether it comes from safety, accessibility options, feature reliability, or another metric.
John Schnoes mentioned that there were benefits to both autonomous and manual driving which a customer may be interested in. Autonomous driving is meant to provide passengers with improved safety, reduced mental loads, and free time for other activities in the car. The benefit of manual driving mode is it allows the driver to experience the pleasure of driving, which is another common customer interest that Nissan received as feedback. In its intelligent mobility development, one of Nissan’s goals is to find a solution that can smoothly shift back and forth from autonomous to manual driving modes such that customers can enjoy both modes.
Unveiled in 2015, the IDS Concept represents Nissan’s vision of intelligent mobility. Utilizing a fully autonomous platform, the IDS Concept will begin on-road testing in Japan in the summer of 2016, with expanded autonomous features such as lane-changing capability to be tested in the coming years. The interior of the IDS Concept reflects an emphasis on the passenger, featuring a flexible arrangement that allows passengers to face each other with rotating seats, along with a multitude of screens to convey information from the car to the passenger. The IDS Concept also features systems to communicate with people outside the car, including a windshield that can display messages to pedestrians and other vehicles.
Nissan IDS Concept
Technology which supplements autonomous driving
GNSS as a potential technology for autonomous driving
|Session: GNSS-based Vehicle Positioning for Autonomous Vehicles|
|U-blox||Brad Sherrad||Senior Vice President|
One of the main points that Brad Sherrad stated in the presentation titled, “GNSS-based (Global Navigation Satellite System) Vehicle Positioning for Autonomous Vehicles” was that as vehicles reach higher levels of autonomy, the vehicle becomes more reliant on autonomous systems for travel. There is a critical point in the progression of autonomy levels to where assistance becomes absolute dependence. Once dependence is required, the vehicle will need information gathering systems such as sensors and GNSS. For a system to be used in a fully autonomous vehicle, it needs to satisfy three requirements:
- Available for use at anytime
- Available for use under any condition
- Extremely, if not completely, reliable
Currently, a combination of technologies, including GNSS, is necessary to achieve all three conditions in positioning, as there is no single technology which satisfies all three requirements concurrently.
Notably, GNSS is the only system currently available that can provide information about absolute position, that is, the position of something based on a fixed reference frame. This allows an autonomous system to calculate additional information such as absolute velocity and heading. GNSS can be contrasted to systems such as cameras or radar, which can only provide relative position, that is, the position of one object based on the position of another. For example, GNSS can provide the location of a vehicle in terms of latitude and longitude, while a vehicle’s radar system can provide the information that there is a pedestrian 100 meters in front of the vehicle. GNSS provides the benefits of removing limitations from only using relative positioning systems, augmenting the information gained from relative positioning systems, and/or reduce dependence on the relative positioning systems themselves. Drawbacks to GNSS include its inoperability indoors, in tunnels, or under dense foliage, thus necessitating the need for other technologies. A use case of GNSS can be seen in the image below, where the system can identify a vehicle changing lanes on a freeway. A table of requirements for GNSS use in autonomous driving is also listed below.
Example of GNSS system identifying lane changes
List of GNSS requirements needed for autonomous driving
Vehicle map development from a software-based focus
|Session: Taking Control of the Map: ADAS, Navigation, and Video Game Tech|
In the presentation titled, “Taking Control of the Map: ADAS, Navigation, and Video Game Tech”, Eric Gundersen gave a brief overview of Mapbox and how it was positioning itself in the autonomous vehicle market with its new offering, Mapbox Drive. As a software-based company, Mapbox is hoping to take its collected experience in the mobile and software markets and apply it to the automotive field. Mapbox provides both an application programming interface (API) and a software development kit (SDK) for its customers. The SDK allows Mapbox to collect data from devices using Mapbox apps, which act as a network of real-time sensors and enhances the data that the company uses from the open-source Open Street Map. Mapbox aggregates and anonymizes this data to provide additional functionality to its users, providing specific map layers such as directionality and real-time traffic updates.
A key component to the Mapbox software is the customizations available for its maps. Developers can customize both the appearance and functionality of maps, giving automakers a high degree of control of what and how information is displayed to their customers. For example, one company could tune their map to be more applicable to semiautonomous driving with lane centerlines, while another focuses on turn-by-turn navigation. An automaker focusing on electric vehicles could display where charging stations are located, while a manufacturer with a heavy global presence can highlight dealership locations. In addition, the maps are displayed with a frame rate of 60 frames per second. Mapbox Drive was unveiled on June 1. During the presentation, Eric Gundersen also announced that Mapbox had signed a major agreement with OEM that would be revealed in the near future.
Perspective on autonomous vehicles from outside the automotive market
Government views and thoughts toward autonomous vehicle development
|Session: Guidelines and Regulations: Catching Up With the Pace of Autonomous Technology|
|University of South Carolina, School of Law||Moderator: Bryant Walker Smith||Assistant Professor|
|National Highway Traffic Safety Administration||Dr. Mark Rosekind||Senior Administrator|
|Nevada Department of Motor Vehicles||Jude Hurin||DMV Administrator|
|Michigan Department of State||James Fackler||Assistant Administrator for the Customer Services Administration|
The panel session titled, “Guidelines and Regulations: Catching Up With the Pace of Autonomous Technology” provided insight into the perspectives of various government entities during the development of autonomous driving technology. The session opened with each of the three panelists giving a brief overview of recent activities which they were involved with. Based on Secretary Anthony Foxx’s announcements in January, NHTSA will release four documents in July 2016, each detailing a specific aspect of autonomous development. The documents will focus on guidance for the deployment and operation of autonomous vehicles, a model state policy, structuring interpretations and exemptions, and identifying new tools and authorities to accelerate autonomous technology.
As a testing-only state, Michigan is looking to expand testing to include driverless vehicles and eventually move towards open deployment. Other items being studied include platooning for commercial and non-commercial vehicles, and a ride-sharing project managed by automotive manufacturers to increase public exposure to autonomous vehicles. One project that Nevada is working on is a center for advanced mobility, where agencies across all levels of government work together to ensure that Nevada maintains its leadership in autonomous regulations. Nevada is also revising its testing and consumer deployment regulations, as well as looking into platooning and autonomous taxis.
One key theme that was brought up repeatedly during the panel session was the importance and need for collaboration between all involved parties. One example of this was the response that Jude Hurin gave when asked if there was anything that he wished he could have done differently, given the benefit of hindsight. Jude Hurin mentioned the mistake of not making sure that all stakeholders were involved after the regulations were made. The involvement of various agencies and entities is important in deploying the technology in a safe manner. Mark Rosekind explained that the federal government is interested in collaborating with everybody, as increasing the amount of collaboration provides more opportunities to discover the best approaches to deploy technology.
Another concept that was raised was the issue of knowing “How safe is safe enough?” Mark Rosekind suggested new metrics and data sources would be needed to measure safety in an autonomous environment. For example, assume a case study where an autonomous vehicle avoids a collision through a hard braking technique. Data recording devices will not show the event as an “avoided crash” but rather as an unsafe “hard braking” event. He also mentioned that equivalency in safety for autonomous vehicles is not enough, and that a higher standard is needed to promote safety as a benefit of autonomous vehicles. Jude Hurin suggested that a slow, step-by-step procedure would be most beneficial to improving safety. Despite the precautions, James Fackler mentioned that the growth of autonomous technology has led to a number of new companies and organizations entering the automotive industry for autonomous testing purposes, creating an atmosphere of anticipation and excitement.
Considerations from experience in rocket development for the automotive industry
|Session: What Did the Rocket Say to the Sedan? Lessons from Space for the Autonomous Car|
|Wind River||Marques McCammon||General Manager, Connected Vehicle Solutions|
Marques McCammon’s presentation, titled “What Did the Rocket Say to the Sedan? Lessons from Space for the Autonomous Car” focused on generalizations gained from experience in the aerospace field that could be applied to the development of autonomous vehicles. The principal theme of the presentation was focusing on the “How” instead of the “What”. Development should be focused on finding solutions such as how to mitigate risk, how to manage complexity, or how to maximize value. A key concept is to avoid having too large of a project scope. As the complexity involved with automated driving is substantial, trying to accomplish too many goals at once reduces a project’s rate of success, or leads to delays and increased costs. Since automated driving is a new technology that the general public has a degree of caution towards, any technological setbacks will hurt adoption and perception of autonomous vehicles.
Another suggestion was that companies should focus on where value is created in the system. Marques McCammon provided the analogy of purchasing a house, where house buyers would be more focused on rooms and sizes rather than the foundation of the building. In the context of the automotive industry, value is created based on user experience. Another generalization emphasized the importance of a system’s architecture. Defining, understanding and constant revisiting of a system’s architecture throughout the development process will generally lead to a project’s success. Marques McCammon gave a case study in the development of the Dodge Neon and the PT Cruiser where this did not take place. Chrysler had decided to build the PT Cruiser off the same platform as the Neon. However, as development continued, the PT Cruiser’s architecture evolved independently into its own platform, therefore failing the company’s initial goals. In regards to autonomous development, Marques McCammon brought up an architecture based on cognitive integration, connecting content with context along with a mechanism to enable system learning.
<Automotive Industry Portal MarkLines>