奔向2021：自动驾驶汽车发展现状和重要驱动者（英文版）.pdf

By Brian Solis, Principal Analyst  at Altimeter, a ProphetCompanyUpdated: Feb 27, 2017THE RACE TO 2021:THE STATE OF AUTONOMOUS  VEHICLES AND A “WHOS WHO”  OF INDUSTRY DRIVERSState of Autonomous Vehicles: Executive SummaryFor this market profile report, we examined the autonomous offerings of 26 auto manufacturers and 50 hardware and software providers, uncovering the following insights and trends: Semi-autonomous vehicles are the stepping stone to fully autonomous vehicles, with most car manufacturers and technology companies taking the lead of Te s la an d  offe r in g  fe at u r e s  s u c h  as  s e lf-parking, adaptive cruise control, emergency braking, and semi-hands off driving within highway/interstate conditions. Semi-autonomous features help consumers become comfortable with the idea of robots taking the wheel. Startups and technology leaders are driving the accelerated innovation in autonomous technology, forcing incumbents to partner, acquire, or ramp up R&D to compete (e.g. BMW and Baidu; Fiat Chrysler and Waymo, an Alphabet company; and GM and Lyft). Toyota, Intel, and Mercedes-Benz have dedicated business units. Automakers are essentially getting into the software/hardware and utility business as future profits will depend less on manufacturing, selling and financing automobiles and more on monetizing driving. Luxury car manufacturers such as Jaguar Land Rover, Maserati, and Porsche are looking toward a future where the driver still has an experiential role in autonomous vehicles. Cars will become platforms for passenger experience, creating a new canvas for cockpits and space. In addition to data science, social science is also becoming prevalent in autonomous development as companies such as Nissan and Audi take an anthropological approach to teach self-driving cars to act more human in their control and on-road actions (e.g. honking, signaling other people or vehicles, moving closer to lane marketing before switching lanes). Large automakers, tech companies, and mobility providers are taking on "acqui-hiring" strategies, acquiring startups (or partnering with them) to absorb not only their technology but also their talent. We'll see a surge in tech jobs in the autonomous industry in coming years on the programming and development side. And, in response to all the acquisitions and partnerships, we're seeing other startups approach the market from a "plug and play" perspective; as in, their technology (hardware or software) can be easily retrofit into existing models of cars. This solidifies their position in the market without the need for a buyout, as their model works more like a supplier. With the latest round of updates from CES, there is more focus on intelligent vehicles (e.g. Toyota, AutoLiv, Ford's concept vision video). These self-driving cars can think on their own, communicate with other vehicles (V2V) through deep learning and AI. This means that all software that exists may run into compatibility issues when asked to communicate with other systems within a vehicle. This is similar to the interoperability conundrum the industry witnessed around IoT. However, when solved for, V2V opportunities abound in improving traffic experience, reducing accidents, and new value-added in-car design components. Mapping software has emerged as its own category among technology providers in the autonomous space, as 3D terrain mapping is a critical component to the effectiveness and safety of self-driving cars as they navigate their environments. Though most autonomous leaders are testing prototypes on the road in Silicon Valley, Austin, Pittsburgh and other metropoles, Honda and Hyundai are focusing their autonomous vehicle testing in large-scale controlled environments that allow for pedestrian situation testing as well as high-speed loops and simulated cityscapes. Progressive automakers are repositioning their future foci away from just “making cars” to becoming mobility services and sharing companies, i.e. BMW, Daimler, Fiat-Chrysler, Ford, GM, Mercedes, Nissan, Tesla, VW, et al. To this trend, ride/hail companies are also attracting investments from traditional automakers to develop next-generation autonomous services. Toyota invested an undisclosed amount of money in Uber. Volkswagen AG invested $300 million in Gett, an Uber and Lyft competitor. General Motors Co. acquired a stake in Lyft. Autonomous vehicles, to varying extents, are already operating in vertical applications such as farming, warehousing/inventory management, and construction. Uber, Tesla, and Mercedes-Benz are also experimenting with other applications of self-driving technologies in city busses and semi-trucks. Carmakers will also become data companies, borrowing cues from Apple, Google, and Facebook to convert data into insights and customize consumer services to deliver value-added experiences. Companies such as BMW iVentures and Toyota Research Institute are already partnering with data startups such as Nauto to share driver data as a means of more rapidly improving autonomous vehicle systems.Update Feb 27, 2017: Additional Executive Insights There is unprecedented collaboration between would-be competitors to accelerate the development of self-driving cars. These collaborations and individual efforts are investing in scaling self-driving technology to consumer markets by reducing technology costs. For example, Waymos initial implementation of LIDAR came at a cost of $80,000 per car, but recent efforts are seeking to bring those costs down to $50 at scale. LIDAR was originally touted as the go-to tech for radar mapping in self-driving cars, but now we're seeing more startups emerge that are using other vision processing technologies with longer range and higher resolution optics for better sensing. This is one reason why LIDAR is decreasing in costs, as better "seeing" technology is emerging in the market. When analyzing automaker and vendor announcements, public policy and infrastructure, and test results, the year 2021 seems a realistic launch date for Level 5 self-driving cars from a technology perspective but not at consumer levels. Initial applications for self-driving cars will be limited to fixed public transit courses, university and business campuses, and inner-city ride/hail services where infrastructure and pedestrian laws have been adapted for safety. Consumers are largely wary of autonomous cars, with many still questioning viability, ethics, and trust. They want to drive a vehicle with semi-autonomous features, and automakers are introducing these features in waves to tame fears and humanize the technologies before fully autonomous vehicle release and adoption is feasible. Self-driving cars are the latest "gold rush" sparked by public tests from the likes of Google and Tesla. Detroit and auto capitals around the world were pushed to accelerate self-driving programs to appear innovative and not lose brand equity.The Five Levels of Autonomous DrivingThe Five Levels of Autonomous DrivingLevel 0: Zero Automation - Driving as UsualA human driver is required to operate the vehicle safely at all times. Driver in full control Eyes on the road Hands-on the wheel Foot on the acceleration pedal/brakeLevel 1: Driver Assisted/Function-Specific -Intelligent features add layer of safety and comfortA human driver is required for all critical functions. The car can alert the driver to conditions, environment and obstructions. It can also offer assisted/smart performance and driving capabilities. Driver in full control Eyes on the road Hands-on the wheel (relief offered in certain modes) Foot on the acceleration pedal/brake (relief offered in certain modes)Vehicle aids driverLevel 2: Partial Automation/Combined Autonomous Functions - Key automated capabilities become standard but driver still in controlAt least two simultaneous autonomous tasks become are managed by the vehicle in specific scenarios. Driver in control Eyes on the road Hands-on the wheel or ready to be on the wheel in cruise control mode Foot on the acceleration pedal/brake or ready in cruise control modeVehicle in partial, temporary control (fixed scenarios)Level 3 : Conditional Automation/Limited Self-Driving - The car becomes a co-pilotThe vehicle manages most safety-critical driving functions in known (mapped) environmental conditions. A human driver is still present and expected to manage vehicle operation. Driver in partial control Eyes temporarily off the road but still observant Hands-off the wheel in specific scenarios but at the ready Foot off the acceleration pedal/brake but at the readyVehicle in conditional control (known environments)Level 4High Automation - Capable of performing all safety-critical driving functions while monitoring environments/conditions in defined use casesPer NHTSA, this is full self-driving automation. Per SAE, Self-driving is fully possible in most road conditions and environments without need of human intervention. A functional driver cockpit is still in place (steering wheel, brake/acceleration pedal, etc.) Driver becomes passenger but can assume control Eyes off the road Hands-off the wheel Foot off the acceleration pedal/brake Vehicle in control (once input is provided and in most situations)Level 5Fully Autonomous - Vehicle is completely driverlessNo level 5 per NHTSA. Per SAE, full-time automated driving in all conditions without a human driver. These vehicles will not feature driving equipment and will no longer look like the vehicles of the past. No driverVehicle in controlThats the year auto manufacturers have promised fully autonomous  vehicles on the road. Depending on who you follow, 2021 (or 2020, in some  cases) is either overly optimistic or cautiously conservative. Either way, its clear  that startups and incumbents are racing toward a self-driving future with press  and media documenting every mile-marker. What remains unclear, though, is whenand to what extentautonomous vehicles will arrive on the roads  and how consumers and commercial industries will adopt progressing levels  of autonomy over time. With the increasing level of activity in the autonomous space, and more companies, products, and partnerships expected to enter the  fray, mapping the ecosystem proves a complex and ongoing commitment.In our attempt to keep pace with the frequency of press announcements in the  space, it became clear that information is distributed and rarely accessible in a  centralized, public manner.This report serves as a comprehensive primer for those monitoring the state  of autonomous vehicle development. It offers an introduction to key players,  organized by industry sector, to outline initial and evolving applications for  consumer markets and verticals. It will be updated at key intervals when  significant announcements are made or milestones achieved.Get Out of My Dreams and Into My Self-Driving CarSelf-driving cars have always stoked our imaginations. From movies, to cartoons, to  advertisements, autonomous vehicles portray a life only possible in science fiction or a dream to  be realized off in the distant future. Though our “Jetsons” or “Minority Report” moment is still  forthcoming, autonomous capabilities are arriving in increments. With each new invention,science  fiction is transformed into reality, one chapter at atime.Tod ay,  as p e c t s  of self-driving cars are making their way into new models, introducingintelligent,  driver-assisted features that are slowly bridging the gap between semi- and fully autonomous  abilities. This is years in the making ofcourse.The race toward self-driving vehicles is as much about technology as it is about infrastructure  and support. Beyond sensors, cameras, processing, etc., everyday things that drivers take for  granted today also have to advance. From lanes and lights, to signs and obstructions, to rules  and regulation, government, cities and manufacturers must also improve to protect people  inside and outside the car. All of this needs to rapidly evolve to keep up with autonomous  technologies andcapabilities.Like any new technology, prices initially appeal to specialized industries such as transportation  and deliveries, as well as individual consumers willing to spend big. Eventually, semi- and fully  autonomous vehicles will be offered in shared and owned formats for almost everyone toenjoy.Autonomous technology will ultimately change the entire automotive industry and supporting  ecosystems and supply chains. Lets consider ownership, financing, and insurance? Data showsthat on average, cars are parked 95% of the time.1With such downtime, traditional ownership could be  considered obsolete. Shared or rental models now become more economically feasible. Users could  simply summon the car on-demand or based on algorithmic patterns of need. As such, automotive  financing may now shift from programs for individual ownership to shared ownership models. Or, if  companies like Uber successfully transition to the self-driving economy, service providers would assume  vehicle leasing/financing programs to offer personalized autonomous driving services.As new ownership scenarios play out, insurance and maintenance industries will also need to  productize new solutions to various stakeholders. Cars will be able to drivethemselves to dealership services. And, because many self-driving cars will be catering to  different stakeholders, insurance companies will stabilize potential losses in  traditional income with modern, scenario-basedpremiums.Open the door, buckle up, and enjoy the ride as we explore the  autonomous ecosystem. Theres a lot of to see, so think of each  section as a vista point. And, dont worry, there are rest stops  along theway.“To finishthe moment,tofindthe journeysendinevery  step of the road, to live the greatest number of good  hours, iswisdom.”- Ralph Waldo Emerson1 shoup.bol.ucla.edu/PayAsYouPark.htmAutonomous Driving Applications v1.0VEHICLE MANUFACTURERSAudi has revealed a series of autonomous vehicle prototypes,  including consumer-oriented test vehicles, based on its A7  and RS7 models. Audi is working on plans to bring its “AudiPiloted Driving” to market in its next-generation flagship model  A8. Initially, the A8 will have the ability to park itself and drive  autonomously up to 37 mph. Audi is also working with NVIDIA, supplier of AI driving technology, on expediting a fully autonomous vehicle program to hit the market by 2020.2Audi announced in July 2016 a dedicated subsidiary devoted to  developing autonomous driving tech called SDS Company.Similar to Nissans anthropological approach to making  autonomous driving more human, Audi has been teaching its  robotic vehicles to drive more like humans in an effort to make  them safer on the roads. For example, the car will move closer to  lane markings before signaling