AI News, Making driverless cars change lanes more like human drivers do

Making driverless cars change lanes more like human drivers do

In the field of self-driving cars, algorithms for controlling lane changes are an important topic of study.

But most existing lane-change algorithms have one of two drawbacks: Either they rely on detailed statistical models of the driving environment, which are difficult to assemble and too complex to analyze on the fly;

“The optimization solution will ensure navigation with lane changes that can model an entire range of driving styles, from conservative to aggressive, with safety guarantees,” says Rus, who is the director of CSAIL.

For any given method of computing buffer zones, algorithm designers must prove that it guarantees collision avoidance, within the context of the mathematical model used to describe traffic patterns.

With the MIT researchers’ system, if the default buffer zones are leading to performance that’s far worse than a human driver’s, the system will compute new buffer zones on the fly — complete with proof of collision avoidance.

“The autonomous vehicles were not in direct communication but ran the proposed algorithm in parallel without conflict or collisions,” explains Pierson. “Each car used a different risk threshold that produced a different driving style, allowing us to create conservative and aggressive drivers.

Using the static, precomputed buffer zones would only allow for conservative driving, whereas our dynamic algorithm allows for a broader range of driving styles.” This project was supported, in part, by the Toyota Research Institute and the Office of Naval Research.

New Algorithm Allows Autonomous Cars to Change Lanes Around Other Cars in Real Time

While this sounds easy enough, these models are difficult to assemble and can be too complex for the car to quickly analyze in a real-life driving situation.

These algorithms can also be too simple and force the car to make impractical and often apprehensive decisions, which can lead to the car never changing lanes.

The autonomous cars using this new system maintain collision avoidance even though the buffer zones are created in real time.

The new system creates a new logistic function in real time based on the estimated direction and velocity.

The skewed distribution is the key to creating a new buffer zone developed while the car is driving in real time.

'The autonomous vehicles were not in direct communication but ran the proposed algorithm in parallel without conflict or collisions,' explains Alyssa Pierson, a postdoc at CSAIL and first author on the new paper, 'Each car used a different risk threshold that produced a different driving style, allowing us to create conservative and aggressive drivers.

Using the static, precomputed buffer zones would only allow for conservative driving, whereas our dynamic algorithm allows for a broader range of driving styles.'

Connected car report 2016: Opportunities, risk, and turmoil on the road to autonomous vehicles

The race to build the fully connected car, and ultimately the completely autonomous vehicle, is already under way.

based on extensive market research, interviews with auto industry experts, and engagement with auto manufacturers, suppliers, and technology companies across the globe—

Although connected services will generate sales of US$155 billion, most of this value will be offset by falling sales from legacy features such as navigation, entertainment, and safety systems.

On the supply side, by 2030, profits available to traditional automakers and suppliers may drop from 70 percent to less than 50 percent of the industry total.

The report is divided into seven sections, each of which focuses on one key question about the opportunities and risks to be found in the industry’s business models, ecosystem, market growth, geographic distribution, and technologies involved in developing the connected car: Automakers, suppliers, and technology companies are beginning to jockey for position.

Recent innovations allow automobiles to monitor and adjust their position on the highway, alerting drivers if they are drifting out of their lane, and slowing down if they get too close to the car in front of them.

The vehicle of the future is already taking shape in a variety of forms, although it is unlikely to reach full fruition on public streets and highways for 10 to 20 years.

But first, it will be useful to put it all in context, and look at the market shifts and structural changes that are underpinning the current and future development of the connected car and autonomous vehicle.

These trends explain why automakers have been investing so heavily in connected technologies, new ride-sharing services, and other transportation options, including such deals as Toyota’s investment in Uber, VW’s in Gett, and GM’s in Lyft.

And revenues from ride-sharing, robofleet, and similar sectors will grow even more rapidly, along with revenues from pure digital services such as onboard entertainment and location-based information providers.

Profits from new cars will decline as the industry shifts to less differentiated, low-cost vehicles such as robo-taxis, as robofleets put pricing pressure on the automakers, and as the cost of the technology in cars rises.

And shared mobility and digital services will capture a much larger portion of overall profits, thanks to significant growth in these businesses, and the healthy margins they can achieve.

We expect more electric vehicles to be sold, and the car to take on more specialized forms, including high-end long-distance vehicles, low-cost/high-volume urban pods, and robo-taxis and other ride-sharing vehicles.

The answer is not necessarily to pump more investment into connected car or autonomous driving technologies, but to invest more thoughtfully: to recognize where your company’s strengths fit with the new technologies, and how to build the capabilities to differentiate your company and stand out in the new technological environment.

Most of them now offer dashboard screens that allow drivers to manage various in-car functions and monitor the car’s status through a digital interface, but otherwise, cars have not evolved all that much from the models of the 20th century.

Five of the most likely value creation levers are as follows, ordered so that those with short-term cash flow potential come first, and those with long-term attractiveness later: Success in these endeavors won’t depend on working out what customers want and then building it.

Rather, it will require a full-on rethinking of how OEMs, suppliers, and technology companies operate, separately and together, within the emerging business ecosystem for connected car development.

This will demand that companies reassess the strategies they use to create value, the capabilities needed to carry out those strategies, and even the corporate cultures that underpinned their traditional, pre-digital ways of doing business.

In the coming years, OEMs will likely sell their cars, and try to capture added value, with three primary connected package options: safety (including driver assistance, lane management, and the like), autonomous driving (including adaptive cruise control and self-parking, among others), and connected car features and services, including vehicle management, consumer-oriented, and commercial applications.

The great risk faced by all OEMs as they move further into the world of connected cars is that they could be outpaced by third-party providers, which can bring the same or similar services to market, either at a significantly lower cost or for free, through an entirely different monetization model—

In either case, the OEMs will have a hard time staying competitive against large digital players, which do not carry the cost of physical assets and the many burdens that come with those assets.

As a result, many OEMs may eventually decide to focus on their old core business, producing commodity cars through which others capture the premium value associated with connected car services.

Driven by the huge demand for the new digital technologies associated with connected cars, an exploding list of companies from outside the traditional automotive supply base—

Not to be outdone, the industry’s OEMs and traditional suppliers are working feverishly to expand their own access to the new technology and talent needed to compete, even as they invest heavily in downstream businesses such as ride sharing.

Successful strategies will include the following measures: The fast growth of new players entering the fray from outside the automotive arena is already beginning to transform the structure of the auto industry.

Making more intelligent, connected, and ultimately autonomous vehicles a reality requires a level of hardware and software technology innovation and adoption never before seen, and much of it has its roots in other industries.

Established players in areas such as telecommunications (think AT&T), IT and software (Cisco Systems), and consumer electronics (Apple) are actively moving to apply their capabilities to the automotive world, leveraging the huge scale and learning effects from much larger consumer and commercial markets to new automotive technologies.

In late 2015, Daimler, Audi, and BMW combined forces to buy Nokia’s precision mapping division, called Here, partly to prevent the service from falling into the hands of a potential future competitor such as Google or Apple.

Leveraging new players will likely be especially important in the areas of software development and artificial intelligence, given the rapid innovation and development processes and trial-and-error mentality so critical to success in both domains.

Software development and coordination are similarly complex for the artificial intelligence features in ADAS and HMI systems that enable vehicles to learn the preferences and styles of different drivers and eventually operate on their own.

GM, for example, has been reducing outsourcing contracts and instead has hired more than 8,000 software developers, while Bosch is looking to employ 14,000 software engineers in 2016 alone to develop features for the connected vehicle as part of its push into the Internet of Things.

How companies fare in the race to provide these features will largely be a function of whether they can build, buy, or partner for the distinct technologies and capabilities on which each feature depends (see Exhibit 8).

Technology, hardware, and software players are entering the market and are helping to accelerate these developments, either by transferring their technologies from other industries to automotive or by starting new companies to focus on specific new technologies for future vehicle applications.

Its powerful Tegra X1 chip can process images from various data sources, such as cameras, radar, and laser imaging, and enable machine learning for automotive systems.

This year, for example, automotive supplier Continental acquired ASC’s Hi-Res 3D Flash Lidar business and its technology for using laser beams to measure distances to cars and other objects on the road, expanding its portfolio of sensor technologies.

Early in 2016, GM acquired Cruise Automation, a maker of self-driving technology, to bring in-house its ADAS-specific software capabilities and to tap into Cruise’s deep software talent and rapid development capabilities.

The deal adds TRW’s capabilities in radar and vision systems, safety-oriented onboard computers, and electronic power steering to ZF’s current portfolio, enabling it to offer OEMs more sophisticated, integrated ADAS systems.

Collaborating closely, the team developed a modular infotainment system, decoupling software from hardware development and cutting down the development time for a new system from as long as seven years to just one.

The deals have helped make Harman arguably the industry’s most successful provider of infotainment systems and related services, and have given it access to over-the-air update and cybersecurity technologies.

Their long history of analyzing driver behavior gives them a distinct advantage, and they understand the rules by which regulators such as the U.S. National Highway Transportation Safety Administration try to ensure safe, undistracted driving.

Meanwhile, auto suppliers and OEMs alike are actively integrating more specific HMI technologies from non-automotive companies such as Nuance (voice control), Immersion (interactive touch features, or haptics), and MyScript (handwriting recognition, which is useful for decoding finger movements).

Among the traditional suppliers is Valeo, which acquired Germany’s Peiker in 2015 to gain access to its onboard telematics and mobile connectivity technology in order to build secure, high-speed connectivity solutions.

On top of the communications infrastructure, companies are developing a variety of services that can aid in the safety and management of the vehicle itself, such as remote vehicle diagnostics, cybersecurity, over-the-air system updates, fleet management, and usage-based insurance.

In 2015, BMW and Pivotal formed a partnership to provide the automaker with big data and predictive analytics capabilities, allowing BMW to better understand the driver experience and to gain valuable insights into vehicle performance—

They include smartphone-based services such as music streaming, e-commerce, social media, integration with smart homes, access to city services like traffic management, and transportation services including ride hailing and car sharing.

And in 2016, GM invested $500 million in Lyft, a ride-sharing service that competes with Uber, in hopes of opening up new vehicle sales channels and eventually driverless taxis.

Finally, among the main objectives of Ford’s strategic investment in Pivotal was the acceleration of its cloud-based software development, in hopes of delivering innovations in the area of mobility services to customers more quickly.

The roles throughout the automotive supply chain are blurring and recombining, and it is critical that every player understand where it stands, and where it should be heading, if it is to capture its fair share of value in this swiftly changing market.

No matter how the government’s political and economic policies evolve, per-capita disposable income is likely to increase for both urban and rural residents, and the country’s middle-class consumer base—

Sales of light vehicles through May 2016 were on pace with the economy as a whole, up 6.9 percent to 10.2 million units, thanks in no small part to the tax stimulus that will continue through 2016.

Indeed, thanks to a virtuous circle of connected consumers, supportive government, and advanced technologies, China may take the lead in the worldwide race to build connected cars.

In fact, several surveys have shown that when making purchasing decisions, Chinese customers are more concerned about a car’s in-car technologies than its design or performance, and would be willing to change brands for better connectivity.

More than 75 percent of Chinese customers would be willing to spend more for safety features, and more than 60 percent would pay more for vehicle management features that track usage, run diagnostics, and record accident data.

These consumers rank safety-related features such as collision prevention, danger warning, and emergency calling highest on the list of connected car offerings they would like, followed by such features as infotainment, navigation, eCall, bCall, and vehicle status and maintenance, according to the GfK Insights Blog.

In one statement, the Ministry of Industry and Information Technology announced these goals for intelligent and connected cars by 2025: reducing traffic accidents by more than 30 percent, setting safe autonomous driving speeds of 120 kilometers per hour, lowering energy consumption by 10 percent, and reducing emissions by more than 20 percent.

As a result of its efforts, by 2030, Chinese companies are expected to control 80 percent of the domestic market for vehicle entertainment modules and perhaps 100 percent of the market for satellite navigation systems.

Thanks in part to this government support, Chinese OEMs, traditional suppliers, and technology companies now entering the market are poised for real growth in, and even potential dominance of, the market for connected cars and related systems and packages.

Baidu is also working on a telematics service for cars, called MyCar, that would monitor car- and traffic-related data, which will also aid the company in its ongoing effort to develop an autonomous vehicle.

In addition, it offers three LED screens and space for as many as four detachable 360-degree cameras to record video and take photos, a smart rearview mirror, support for voice controls, and an onboard “intelligent”

The range of other partnerships illustrates China’s thriving connected car market: As advanced as China’s efforts have been in the connected car space, the country has yet to progress very far on the autonomous vehicle front.

That’s in part because China’s driving conditions are complex in terms of traffic, road conditions, and driving behavior, compared with countries like Sweden and the U.S., where companies are actively testing these systems.

Although China will ultimately end up with a system tailored to its own realities, these complexities will undoubtedly slow its progress, which depends in part on adopting insights and technologies from elsewhere.

To achieve their goals, these companies need to become more active in this space, focusing on offering more connected car features to their domestic customers and improving their innovation capabilities internally or through partnerships, as SAIC is doing with Alibaba.

Indeed, if this trend continues, China’s tech companies, already the first movers in the country’s highly innovative market, may very well begin competing successfully in global automotive markets as well.

It would be difficult for them to match the quality, reliability, and safety standards the OEMs have already achieved, while fuel consumption, the development of alternative and renewable fuels, and environmental issues will also be barriers.”

In response, outside companies should take advantage of China’s increasingly innovative technology environment to develop and test their own connected car applications and services, and focus more on strategic partnerships with Chinese tech companies, in order to better meet the needs and demands of the Chinese consumer.

digital functions and services become more sophisticated, hackers are likely to turn their attention to stealing the software code for new functions and offering it to users for free, disturbing the entire business case for the connected car.

Aside from the purely technical issues involved, doing so is a matter of putting together the proper project environment in which the security software needed can be developed, tested, and maintained.

The task of coordinating the security software developed internally with the security efforts of third-party suppliers of systems and services adds considerably to the development challenge.

lack of proper testing procedures, moreover, also affects how companies determine the risks involved in the technology in the connected car, and whether their software has adequately addressed them.

Moreover, these procedures must be standardized as much as possible to ensure that the quality of the software is consistent, and that the tests provide measurable results regarding the risk potential and confidence level that can be compared across different elements of the software.

Once testing is completed, and the security software is implemented in cars, software teams will also need to regularly update the software as potential new threats emerge, and send it out—

no easy task given the technical complexity of the software, the need to coordinate the updating process with third-party suppliers, and the long product life cycles of cars.

not just to keep the car and its growing number of connected services safe from hackers, but to instill the high level of trust needed to keep car buyers coming back for more.

Given how complex the effort to secure the connected car is, and the sheer number of different players whose software is used to keep the car connected, securing these vehicles must be a collaborative effort.

The goal: to build the intelligence and learning capacity into cars that will enable them to know who you are, understand your changing moods, adapt to varying circumstances, and react to new demands and new tasks on the fly.

industrial machines have access to a wide range of sensors and network data that give them a detailed picture of their workplace and enable them to interact smoothly with humans or call for maintenance in advance.

And the AI systems will take into account external data sources such as social media, marketplaces for e-commerce and entertainment, and smart home systems, offering updates and suggestions.

Further in the future, perhaps, your car will learn not just from you and its immediate environment, but through its connections with other people and cars, leading to a kind of swarm intelligence that will exponentially increase its understanding and learning potential, enabling it to combine its sensor data with other cars’

Whether AI and machine learning can become a differentiating capability for Toyota, or any other automaker, will depend on just how far it can push the technology, and whether the advantages and features, which will no doubt be offered first as expensive options in premium cars, will rapidly end up as standard equipment in mass-market cars, built into the list price.

now virtually all cars offer little more than an LED screen through which you control the infotainment, navigation, and climate control systems and which provides a wealth of data about the status of the car—

Yet the graphical user interfaces and control mechanisms of many of these systems have proved woefully counterintuitive, making it difficult, even distracting, for drivers to figure out how to adjust them and access information they need.

A car seat, for instance, might change firmness and texture depending on whether the passenger or driver wants to engage more fully with the navigation process, or sit back and relax, and on whether or not it detects muscle stiffness.

Soon, the car will be able to communicate much more information this way, warning us verbally about a traffic jam ahead and suggesting alternative routes, telling us which hotels in the area have vacancies, or updating us on the score of a current football match.

And once the car is fully autonomous, and fully connected to the surrounding infrastructure, its powerful combination of AI and HMI will let you (as the driver) tell it to drop you off at home, go off by itself to find a parking space—

Although outsourcing some of the car’s computing needs to the cloud on the fly is possible, response must be instantaneous, 100 percent available, and easy to plug in, even when moving between cell nodes.

OEMs, suppliers, and any company providing onboard services will have access to far more information about drivers and passengers, and their behavior, interests, and preferences, which can be used to market to them much more precisely.

And automated customization of fleet cars to the preferences of users will smooth the transition from individual ownership to ride hailing and ride sharing and make the use of robocars more attractive.

Players throughout the global auto industry are working hard to make the vision of the connected car and autonomous vehicles a reality, experimenting, testing, and building the technology that will further connect cars to the world around them, and developing prototypes of cars that can already drive themselves.

The number of accident prevention innovations has increased by more than 500 percent, and advanced driver assistance system [ADAS] innovations by more than 400 percent, over the past eight years.

Bratzel: The traditional OEMs are following an evolutionary track to autonomous driving, by incrementally improving the technology and adding features like vehicle-to-vehicle and vehicle-to-infrastructure communications, to make car travel safer and more comfortable.

Bratzel: These new players that are interested primarily in the concept of mobility, mobility-on-demand, and service-oriented business models will evolve out of the trends of connectivity, alternative power sources, mobility services, and autonomous driving.

In the next 10 years or so, however, a new innovation and business dynamic will arise out of the convergence of these trends, generating completely new customer benefits in the world of mobility and beyond, and creating a high probability of disruptive change.

Algorithmic Accountability: Designing for Safety | Ben Shneiderman || Radcliffe Institute

Vital services such as communications, financial trading, health care, and transportation depend on sophisticated algorithms. Some rely on unpredictable ...

2018 Isaac Asimov Memorial Debate: Artificial Intelligence

Isaac Asimov's famous Three Laws of Robotics might be seen as early safeguards for our reliance on artificial intelligence, but as Alexa guides our homes and ...

The Third Industrial Revolution: A Radical New Sharing Economy

The global economy is in crisis. The exponential exhaustion of natural resources, declining productivity, slow growth, rising unemployment, and steep inequality, ...

Swarms of Aerial Robots - AMNH SciCafe

Autonomous aerial robots, commonly referred to as drones, could soon be used for search and rescue, first response, and precision farming. Join roboticist Vijay ...

C. C. Mei Distinguished Speaker Series: Prof. Serge Hoogendoorn

This video features the lecture of Prof. Serge Hoogendoorn discussing “Making Mobility Smart Again,” on April 23, 2018, CEE, MIT. The MIT CEE Distinguished ...

Yelawolf - Daddy's Lambo

Sign up for updates: Music video by Yelawolf performing Daddy's Lambo. (C) 2011 DGC Records Best of Yelawolf: ..

Digital Transformation: Interview with Bob Noddin, AIG Japan Holdings

"Digital Transformation: Visions of Nations, Companies, and People" is a documentary project by Manuel Stagars

Moral Math of Robots: Can Life and Death Decisions Be Coded?

A self-driving car has a split second to decide whether to turn into oncoming traffic or hit a child who has lost control of her bicycle. An autonomous drone needs ...

Bina Hallman & Steven Eliuk, IBM | IBM Think 2018

Bina Hallman, VP, Offering Management Storage & Software Defined at IBM, and Steven Eliuk, VP, Deep Learning, Global Chief Data Office at IBM, sit down ...

Here be Dragons 2018: Track A

Sea monsters such as the kraken, prister, and rosmarus indicated uncharted territory on elaborate new maps of the world in medieval times. Despite many ...