Safety First

The Lexus LS is a regular series production vehicle that will be driving on regular, open roads. The only difference is the roof mounted pack of sensors including LIDAR, radars, cameras and high precision positioning system. In the car, a Safety Driver, who can intervene and overrule the AD vehicle control system at any time, will be accompanied by an Operator supervising the entire system. Toyota is bringing this car to the open roads, after months of thorough preparation of the vehicle: validating the AD system, training the drivers, analysing the route and engaging with authorities to acquire the necessary approvals.

This research ties in with Toyota’s global vision to create safe and Ever-Better Mobility for All. Toyota sees the relationship between a driver and the car as teammates working together, a mobility approach which it has dubbed ‘Mobility Teammate Concept’. With the aim to realise more safe and accessible mobility, Toyota is working on two different automated driving models in parallel: Guardian and Chauffeur¹. Whilst developing these cutting-edge technologies, Toyota is also popularizing safety features in an affordable way, across the range. Toyota equips all its newly sold cars with advanced driving assistance systems (ADAS), called Toyota Safety Sense and Lexus Safety System+².

European L3Pilot

The automated vehicle will also be used for collecting data as part of Toyota’s involvement in the European “L3Pilot project,” along with 34 other partners, including major car manufacturers, automotive suppliers, research institutes and authorities. The L3Pilot is a four-year European project launched in 2017 and partially funded by the European Commission³. The project paves the way for large-scale AD field tests of around 1,000 drivers in 100 cars across 10 countries in Europe. Within this framework, TME will concentrate on researching customer behaviours and the safe operation of systems in complex and diverse urban environments.

¹ Guardian mode uses technology to constantly monitor the human’s driving task, intervening only when necessary. In Chauffeur mode, the technology takes all responsibility for driving.

² The Lexus LS is equipped with Lexus Safety Sense+A

³ This project has received funding from the European Union’s Horizon 2020 research and innovation programme; grant agreement 723051, Funded under H2020-EU.3.4.

CARLA is hosted on Github, and its development is directed by the CVC at the Universitat Autònoma de Barcelona (UAB) in Spain. The non-profit, open source simulator is for the development, training and validation of automated urban driving systems. It is designed to ensure the stability of automated vehicles in myriad situations that are not always testable in the real world. With flexibility and realism in mind, CARLA offers a multitude of environmental conditions and is designed to be extended and easily modified to fit specific project needs.

The team behind CARLA plans on using TRI’s gift to expand its engineering team and the simulator’s capabilities.

CARLA is another major commitment by TRI to an open source platform. TRI has made significant donations for two consecutive years to the Open Source Robotics Foundation (OSRF), which offers the global robotics community access to open source robotics software and development tools. TRI is also the primary contributor to Drake, an open source C++ toolbox created by the MIT Computer Science and Artificial Intelligence Lab (CSAIL) for analyzing the dynamics of robots and building control systems.

To learn more about CARLA, please visit Carla.org.

About Toyota Research Institute

Toyota Research Institute is a wholly owned subsidiary of Toyota Motor North America under the direction of Dr. Gill Pratt. The company, established in 2015, aims to strengthen Toyota’s research structure and has four initial mandates: 1) enhance the safety of automobiles, 2) increase access to cars to those who otherwise cannot drive, 3) translate Toyota’s expertise in creating products for outdoor mobility into products for indoor mobility, and 4) accelerate scientific discovery by applying techniques from artificial intelligence and machine learning. TRI is based in the United States, with offices in Los Altos, Calif., Cambridge, Mass., and Ann Arbor, Mich. For more information about TRI, please visit tri.global.

Potential outcomes include:

• integrating an AI system for traffic lights (signal) control
• building a platform for interactive data manipulation to monitor and predict traffic behaviour, and to test out planning scenarios
• finding mechanisms for fleet operators and cities to work together, for example by sharing data about congestion or pollution hotspots, and rerouting around the problem before it becomes serious

Media Inquiries

Beth Wood

Press and Communications Manager

The Alan Turing Institute

bwood@turing.ac.uk

(0)20 3862 3390

Julie Ann Burandt

Global Communications Manager

Toyota Mobility Foundation

Info@toyota-mf.org

^{About the Toyota Mobility Foundation}

^{The Toyota Mobility Foundation was established in August 2014 to support the development of a better mobile society. The Foundation aims to support strong mobility systems while eliminating disparities in mobility. It utilizes Toyota’s expertise in technology, safety, and the environment working in partnership with universities, governments, non-profit organization, research institutions and other organizations to address mobility issues around the world. Programs include resolving urban transportation problems, expanding the utilization of personal mobility, and developing solutions for next generation mobility.}

^{About the Alan Turning Institute}

^{The Alan Turing Institute is the UK’s national institute for data science and artificial intelligence.}

^{The Institute is named in honour of Alan Turing, whose pioneering work in theoretical and applied mathematics, engineering and computing is considered to have laid the foundations for modern-day data science and artificial intelligence. The Institute’s goals are to undertake world-class research in data science and artificial intelligence, apply its research to real-world problems, drive economic impact and societal good, lead the training of a new generation of scientists, and shape the public conversation around data. turing.ac.uk}

Kuzumaki: “Toyota wants safe and ever-better mobility for all. We do not want people to worry about the possibility of being involved in an accident. When somebody gets in a car, you would just say “Have fun”, instead of “Be careful.” With Automated Driving we set ourselves the goal to reduce the more than 1.3 million lives currently lost around the world in traffic accidents each year.”

Pratt: “In addition, everyone, and we mean really everyone – including the elderly, and people with physical disabilities – unable to drive today would be able to move safely anywhere they want to go.”

Q. How do you plan to make this happen?

Kuzumaki: “The future mobility we envision is built on the concept that the car will help you remain safe. The car is watching over you, like a teammate would do. That is why we have dubbed our mobility approach ‘Mobility Teammate Concept’.”

Pratt: “Over time, as a car’s artificial intelligence (AI) improves, the car will increasingly learn drivers’ and passengers’ habits and preferences. AI will sometimes follow these preferences, at other times it will warn the driver. Car and driver will evolve together and become inseparable.”

Q. How does that work in practice?

Pratt: “Toyota distinguishes two types of vehicle autonomy: ‘Chauffeur’ and ‘Guardian’. The Chauffeur concept is Automated Driving when the driver doesn’t want to, or cannot drive. In Guardian mode, a human is driving with the support of the system’s advanced driving assistance. What is important to emphasize in our approach is that we leave the choice to the driver. Even when the car will be fully automated, drivers will still be able to enjoy the fun and freedom of driving, as they wish.”

Kuzumaki: “These advanced safety technologies are only one side of the equation. We take a 2 axes approach: as well as developing cutting-edge technologies, we are working to popularize safety features in a more affordable way, across our range. Only when safety technologies are fitted on the full line-up, including entry-level models, can we make a real difference. We are proud that our accident avoidance features today are fitted on 92 percent of our European sales, helping us strive towards zero traffic accidents.”

Q. That sounds promising, but when can we expect to see all of this on the road?

Kuzumaki: “In 2020, we will introduce Highway Teammate. Highway Teammate is Automated Driving on highways: merging onto or exiting highways, overtaking other vehicles and changing lanes. That is, if that is what the driver wants. In line with our philosophy, the system will always collaborate with the driver.”

Q. No Automated Driving on ordinary roads then?

Pratt: “In the early 2020s, we plan to introduce Urban Teammate, which is Automated Driving on ordinary roads. With more advanced AI, Urban Teammate will enable further driving automation.”

Q. What is your stance on collaboration versus competition in this field?

Pratt: “Concerning collaboration, we remain very open. Competition and collaboration need to be carefully managed as we are in an ever-changing environment.”

Kuzumaki: “Indeed, we should make it clear in what fields companies should compete and collaborate. For example high accuracy mapping is an area where information could be shared. Collaboration is important, but so is competition to accelerate innovation.”

Q. How does TRI fit in Toyota’s global R&D structure?

Pratt: “At Toyota, Kaizen or Continuous Improvement, is at the heart of everything we do. It remains the foundation and one of our biggest strengths. However, in these unprecedented times, we need additional approaches. In this particular domain, it’s simply not enough to only take small steps. You have to take leaps, which often results in failure. If you are lucky, after you have the courage to keep on failing again and again, you suddenly succeed. The mission of TRI is to explore spaces and take these leaps for Toyota. In addition to the work we’re doing ourselves we also recently created Toyota AI Ventures, a corporate venture capital fund designed to cultivate and invest in the entrepreneurial spirit of start-ups in the field of Data, AI/AD, Machine Learning, Robotics, …”

Q. How do you aim for the highest level of safety? How safe is safe enough?

Pratt: “Writing the software is relatively easy to do. The hardest part is predicting human behaviour. Will the car turn left or right? Will the child follow the ball? We are building a statistical model based on the data we gather from real human behaviour. From these statistics we create a simulated environment and test our autonomy software in reaction to that. But this is not enough. We simulate many more miles with much worse traffic than actually happens in the real world. Finally, we complement these simulations with physical testing, as the road remains the best test environment. By combining simulation and real world driving, we aim to reach the trillion mile challenge Toyota set itself.”

Kuzumaki: “It is crucial that the vehicle is completely safe on its own. This means without the need for interaction with infrastructure or other vehicles. This ensures the vehicle can safely navigate despite a lack of service caused by a natural disaster, for instance. As these vehicles will have to deal with extreme weather and traffic situations, the system requires a mix of many different sensors that are complementary to one another. Lidar for example struggles with snow.”

Q. To conclude, what are the remaining challenges for the future of mobility?

Kuzumaki: “First, legislation. We need unified regulations if we want to introduce Automated Driving technology safely in various regions. Second, Automated Driving requires transmission of a huge amount of data, and it is important to protect data security and personal data.”

Pratt: “Reaching level 5 autonomy is definitely one of the remaining challenges. AI does not have the sufficient level of human reasoning at the moment. I challenge everyone who says they know when they’ll achieve level 5. Social acceptance is another challenge. Not everyone is ready to embrace AD/AI. However, we start to see a change in mindset. Remember elevators from the past? They used to be ‘manned’ but over time automatic elevators became widely accepted. We must ensure that our products are safe and that automated vehicles perform significantly better than a human driver if we want a mother to trust her children’s life with an automated car.”

For more info on Toyota's Vision for Automated Driving: http://automatedtoyota.com

“Beyond the extraordinary work that U-M is doing broadly in advancing automotive safety research—and in autonomous driving, in particular-- Toyota has deep roots in the Ann Arbor community,” saidPratt, in keynote remarks at the GPU Technology Conference today. “TRI was drawn to Ann Arbor because of the strength of the university; the utility of Mcity and the Mobility Transformation Center which we currently sponsor; the promise of the future American Center for Mobility at Willow Run; and the proximity to, and synergies with, our two well-established Toyota Technical Centers nearby.”

The Toyota Technical Centers have been conducting research in the area of autonomous cars for more than a decade. A group of about 15 team members will transfer to the new TRI-ANN facility when it opens. U-M Professors Ryan Eustice and Edwin Olson are joining TRI-ANN as the area leads for mapping/localization and perception, respectively. Both will be based at the Ann Arbor office, and will retain their U-M faculty positions. 

"Sensor hardware and algorithms are improving at a tremendous pace. TRI researchers will push the frontier even further, with the goal of providing safer vehicles and more helpful robots in the home,"said Olson. Eustice added, “Ann Arbor is a fantastic location for TRI to expand its autonomous driving efforts. We will benefit from Toyota’s existing team and U-M’s research talent and facilities where we can perform extreme-limit testing in a wide variety of environments.” 

Although the focus of each of the three strategically located facilities will be broad, each will feature a different core discipline. TRI-ANN will focus primarily on fully autonomous (chauffeured) driving. TRI-PAL will work on what may be termed “guardian angel” driving, where the driver is always engaged but the vehicle assists as needed. TRI-CAM will dedicate a large portion of its work to simulation and deep learning.

The Toyota Research Institute is an enterprise designed to bridge the gap between fundamental research and product development. With initial funding of $1 billion, it has four initial mandates. 

First, it will strive to enhance the safety of automobiles with the ultimate goal of creating a car that is incapable of causing a crash, regardless of the skill or condition of the driver.

Second, it will work to increase access to cars to those who otherwise cannot drive, including seniors and those with special needs. 

Third, it will help translate Toyota’s expertise in creating products for outdoor mobility into products for indoor mobility; moving people and goods across the country, across town, or across the room.

Finally, TRI will accelerate scientific discovery by applying techniques from artificial intelligence and machine learning, particularly in the area of materials science. Using computation and machine learning, it hopes to accelerate scientific discovery in this area, lowering costs and improving performance of future mobility systems.

Beyond the projects it will engage in with the three universities (more than 30 are already underway at Stanford and MIT), TRI is enthusiastically pursuing collaboration with other automakers, IT companies, suppliers, research labs and universities to jointly develop autonomous technologies.

“Although the industry, including Toyota, has made great strides in the last five years, much of what we have collectively accomplished has been easy, because most driving is easy,” said Pratt. “Where we need autonomy to help most is when the driving is difficult. It’s this hard part that TRI intends to address. Toyota’s goal is safer mobility for all, at any time, in any place, and the tremendous improvements in quality of life that such universal mobility can bring.”

As we age, mobility becomes more challenging and larger segments of society are unable to drive or move freely. Also, the demands on healthcare systems and those who support the physically infirm continue to increase. Toyota believes the opportunities to improve every-day living through artificial intelligence supported technologies are boundless, with significant breakthrough potential for the development of life-saving intelligent vehicles and life-improving robots.

Kiyotaka Ise, TMC Senior Managing Officer and Chief Officer, RandD Group, said: "We're here today to mark the beginning of an unprecedented commitment. We will initially focus on the acceleration of intelligent vehicle technology, with the immediate goal of helping eliminate traffic casualties and the ultimate goal of helping improve quality of life through enhanced mobility and robotics. This partnership, directed by Dr. Pratt, is a great opportunity to work with two leading research teams from two top universities. I am very excited about what this new venture means for Toyota, and I look forward to more announcements in the future."

Dr. Gill Pratt said, "This bold collaboration will address extremely complex mobility challenges using ground breaking artificial intelligence research. I'm thrilled to be a part of the synergies and talent-sharing of Toyota, MIT, and Stanford. Key program areas will be addressed by the two university campuses and Toyota, with combined research targeted at improving the ability of intelligent vehicle technologies to recognize objects around the vehicle in diverse environments, provide elevated judgment of surrounding conditions, and safely collaborate with vehicle occupants, other vehicles, and pedestrians. The joint research will also look at applications of the same technology to human-interactive robotics and information service."

Research at MIT will be led by Professor Daniela Rus, Andrew (1956) and Erna Viterbi Professor of Electrical Engineering and Computer Science and Director of the Institute's Computer Science and Artificial Intelligence Laboratory (CSAIL). "Our team will collaborate with Stanford and Toyota to develop advanced architectures that allow cars to better perceive and navigate their surroundings in order to make safe driving decisions," said Professor Rus. "These efforts will play a major role in helping reduce traffic casualties, and potentially even helping us develop a vehicle incapable of getting into a collision."

Led by Professor Fei-Fei Li, Director of the Stanford Artificial Intelligence Laboratory (SAIL), the Stanford-based research center will collaborate with MIT and Toyota to develop advanced intelligent systems to recognize, understand and act in complex traffic environments. "Building on Stanford's expertise with computer vision, machine learning, large-scale data analysis and human-computer interaction, our team will work to help intelligent vehicles recognize objects in the road, predict behaviors of things and people, and make safe and smart driving decisions under diverse conditions," said Professor Li.

Beyond Toyota's longstanding work on autonomous vehicles and advanced driving support systems, the company has been developing robots for industrial use since the 1970s, and for Partner and Human Support Robot applications since the 2000s. This collaborative effort will open up new avenues for systems and product development across a broad range of mobility applications.

For more details, photogallery, press conference remarks and the video of the announcement, please visit: http://newsroom.toyota.co.jp/en/detail/mail/9233109