self-driving cars

AV Legislation Returns to House, While Some Supporters Say Reconsideration Unlikely

January 7, 2021 by Levinson and Stefani Leave a Comment

The U.S. House of Representatives is considering a bill that would allow car manufacturing companies to distribute up to 100,000 self-driving vehicles. The legislation, called the SELF DRIVE Act, has received little-to-no attention over the last few years. It was passed unanimously in 2017 by the House, but failed within the Senate. Democrats and trial lawyers both worried about potential lawsuits that could arise if someone is injured or killed in an autonomous vehicle.

The measure would give the Secretary of Transportation the ability to make federal motor vehicle rule exemptions regarding a car’s need to have a human operator on board. 25,000 cars per carmaker could be operated initially if a company can show it meets traditional vehicle safety standards.

Following a period of one year, a manufacturer’s exemptions would rise to 50,000 and then to 100,000 in the two following years. States would not be able to limit sales more than the federal statute with any new laws. Right now, a car manufacturer can produce 2,500 self-driving vehicles for testing.

Ohio Representative Bob Latta believes the self-driving measure should resurface in Congress after the House’s major backing in 2017.

“This is a very good bill,” Latta said. “Not only for safety–we had over 37,000 people killed in traffic accidents–but also for the bipartisan work that was done.”

However, Michigan Representative Debbie Dingell said she won’t support the revisited measure, although she was a proponent of the 2017 version, because she doesn’t believe it will reach proper reconsideration in committee or within Congress before the end of the term.

“The Congress should have considered autonomous vehicle legislation, but for a variety of reasons, that has not occurred,” she explained.

Latta’s “desire to see action by our colleagues” is respectable, Dingell said, and agrees that immediate action on the matter of AV is necessary. She also expects 2021 to include many bipartisan discussions on self-driving legislation, and that she has already received Democratic leader commitment to making the issue a priority if the democrats continue their control of the House.

Bringing back the legislation was a “cynical” move for lawmakers who have seemed nonchalant in regards to AV regulation, said Center for Auto Safety executive director, Jason Levine.

“As a reminder, this bill contains no regulatory or testing requirements to improve public confidence in the long-term safety of driverless vehicles,” he said. “Instead, the SELF DRIVE Act protects auto manufacturers and technology companies from legal responsibility when their unregulated products put consumers at risk.”

This news comes shortly after major trucking company Daimler Trucks purchased a majority share in Torc Robotics in efforts to work toward Level 4 autonomous truck development. The companies see the path to AV trucking as a marathon rather than a sprint, and believe they can bring automated trucks to mainstream transportation by 2030.

SAE Level 4 is the name for a truck’s ability to operate without driver assistance or input during certain conditions or in specific areas.

The companies have also been urging Tier 2 suppliers for additional safety tech, and announced that Daimler’s Autonomous Technology Group would build headquarters in Albuquerque. The center will boost testing capabilities and data collection for the use of innovative vehicle technology on public roads.

Torc Robotics CEO Michael Fleming noted that there are still many obstacles ahead, and that safety-focused technology is not quite ready for commercial deployment.

“We have been engaging Tier 1 suppliers with new requirements, trying to push the envelope on radar, lidar, and cameras, and also compute systems at the same time,” he explained. “At the end of the day, when we bring product to market, it has to be cost-effective, meet the performance specifications to address not only safety, but ensuring we can drive efficiently to the destination and deliver goods in a timely fashion…There isn’t, in my opinion, automated vehicle hardware commercially available that meets all three today.”

Still, collaboration is key to pushing new methods of safety technology forward, Fleming said.

“Daimler’s commitment to safety, innovation leadership of truck technology, and foundational knowledge of on-road scenarios that truckers encounter has moved our system faster than we could have done alone as a technology firm.”

Public Trust Key to Future of Autonomous Delivery Vehicles, Experts Say

November 30, 2020 by Levinson and Stefani Leave a Comment

Public trust in automated vehicle technology is the key to the future of e-commerce, experts say. Especially as stay-at-home orders continue and the pandemic limits in-person shopping capabilities, people who rely more heavily than ever on contactless delivery should begin learning what AV tech has to offer.

Head of policy at Nuro–a robotics company that focuses on autonomous vehicle deployment for last-mile deliveries–Matthew Lipka, explained that this public trust is extremely important, and AV systems could in fact be regularly delivering someone’s pizza sooner than we think.

“I think that’s really, really critical,” said Lipka of public confidence around AV tech. “We need to get out there and talk to people and introduce them to the technology. I think delivery technology is a way of building that public trust.”

Alia Verloes of infrastructure and transportation-focused consulting group Steer said community engagement and outreach will help to address any common concerns and also help progress a conversation about how this tech can help in the midst of the pandemic.

“Let’s involve communities at the local level [and] county level to better understand the specific concerns,” said Verloes. “Outreach means many things, but here, it’s about listening.”

The coronavirus pandemic has also brought about a “greater appetite” for delivery services, Verloes noted. She also explained that e-commerce as an industry has seen huge growth in areas like grocery shopping, but it is hard to tell if this pattern is here to stay.

We recently reported on the revealing of the U.S. Department of Transportation’s National Freight Strategic Plan, which aims to boost infrastructure planning and freight movement efficiency. The plan will work toward improving overall freight system safety, as well as the modernization of infrastructure and the development and implementation of innovative technology.

Building a strong infrastructure network involves strong economic competitiveness, according to Transportation Secretary Elaine Chao, and investment in safer and modernized infrastructure and technology could very well include boosted efforts to support autonomous delivery vehicles.

According to the National Freight Strategic Plan, e-commerce shopping habits were already growing extremely quickly even before the pandemic began. Between the fourth quarters of 2018 and 2019, e-commerce sales grew 16.7%, as shown in U.S. Census Bureau data, and has skyrocketed since. The 118-page plan also identified that e-commerce has become a major issue for current freight movement efficiency, as the industry has contributed greatly to the increase of overall truck traffic and curb space competition.

“The rise of e-commerce has disrupted our supply chains and increased demand for last-mile deliveries in areas that are already heavily congested,” said Chao at the announcement of the plan. “This National Freight Strategic Plan will help us invest strategically in our country’s future and turn these challenges into opportunities.”

Verloes explained that AV technology could not only save customers time, but it could also save lives. Shifting toward more regular AV utilization could help commuters avoid 244,000 injury-causing road crashes between 2025 and 2035. Additionally, if AV deployment is implemented during this time period, around 21 billion hours of driving for errand-related purposes would be saved for road travelers.

Because these particular automated vehicles would not accommodate people, they would not need the same hyper-focus on design as autonomous passenger vehicles have, including designs regarding driver and passenger comfort and safety.

“With no driver or passengers to worry about, our vehicle has been engineered from the ground up to keep what’s outside even safer than what’s inside,” said Nuro on its website.

Nuro’s R2 vehicle is designed strictly to transport goods and has been designated to a pilot grocery delivery program in Scottsdale, Arizona. According to Lipka, because the vehicle won’t hold any occupants, it is able to stop and move abruptly without any safety concerns.

Ground-based AV systems also have the ability to transport and carry a variety of items, whereas other mobile technology, like drones, does not. Drones, Lipka said, can deliver anti-venom for snakebites in an emergency, but can’t drive two gallons of milk to a shopper’s curbside.

Lipka also explained that transitioning to automated delivery vehicles in lieu of traditional fleets will need to be a careful and deliberate move, as this technology must operate at “automotive-grade reliability.”

“[They’re] not like scooters, where you’re just going to wake up one morning, and there’s 1,000 delivery robots on your street,” he said.

Uber Vows to Make AV Test Data Public and Boost Safety Efforts After NTSB Backlash

October 22, 2020 by Levinson and Stefani Leave a Comment

After the National Transportation Safety Board blamed Uber Technologies Inc.’s policies for a fatal 2018 crash involving a self-driving vehicle, Uber has promised to prioritize publicizing safety information regarding its self-driving technology efforts.

Uber updated its voluntary safety assessment, sent to the National Highway Traffic Safety Administration on August 28th, with its new pledge. This is the company’s first effort against criticism regarding its autonomous driving program since the NTSB first published its beliefs about the first fatal pedestrian accident involving a self-driving vehicle–one of Uber’s–in Tempe, Arizona in 2018.

“We support the idea of transparency and making the public understand what we do,” said Uber’s Advanced Technologies Group head of safety, Nat Beuse. This new voluntary safety assessment is a “complete update” regarding what Uber originally told regulators in 2018.

The March 18th, 2018 incident involved an inattentive Uber safety operator in an autonomous vehicle that hit and killed a 49-year-old pedestrian crossing a dimly-lit roadway. The vehicle was in self-drive mode while the safety operator sat behind the wheel.

This accident caused nation-wide backlash against self-driving cars, although police say the vehicle was operating for testing purposes only.

The NTSB declared that the driver had failed to act safely while distracted by his or her cellphone, and that Uber was at fault for having subpar safety risk assessment procedures, inadequate vehicle operator oversight, and a lack of any mechanism addressing complacency by operators.

Cell phone distraction or otherwise tuning out has been considered a major issue surrounding AV technology, and more researchers are studying just how much advanced driving systems create worse human drivers. Still, many AV supporters believe any current issues will be solved once self-driving tech improves to the point where, hopefully, car crashes will be eliminated.

In its assessment, Uber points out current “enhancements,” such as a new “Safety Case Framework” it claims will allow for open-sourced peer reviews. Additionally, the company said new internal safety management regulations and an independent Safety and Responsibility Advisory Board will be put in place.

Still, safety advocates have spoken against the Trump Administration’s implementation of voluntary approaches to self-driving vehicle regulation, saying voluntary reports are more like marketing brochures instead of formal regulatory filing submissions.

After Uber’s fatal incident, the legislation being considered in Washington to boost the number of autonomous vehicles manufacturers could produce began being heavily debated. All self-driving car testing was suspended by Uber for four months, and its Arizona driverless testing program was also shut down, causing the layoffs of 300 employees.

Currently, 23 various companies have made public their own self-driving safety assessments, including Apple Inc., Ford Motor Co, General Motors Co., Lyft Inc., Mercedes-Benz AG, Toyota Motor Corp., and Waymo. Uber is one of a few that have begun updating voluntary assessments, Beuse noted.

Advocates of consumer safety are using this example to push for stricter self-driving car regulations and more frequent consumer-focused safety assessment reports. Many have also criticized government agencies for being too lenient on the firms working to improve vehicle autonomy, and on voluntary reporting itself.

“It’s nice that Uber has decided this is the right time to update its so-called report, but a consumer-focused agency would have long ago mandated all driverless vehicle manufacturers regularly submit useful safety details regarding their public road tests,” said Center for Auto Safety executive director, Jason Levine.

The major problem with autonomous vehicle testing is weak federal oversight, said Ensar Becic, an NTSB project manager. He explained that even a regulated, basic self-driving vehicle test, “be it an obstacle course, a perceptual test, or tangible requirements such as testing for miles or adherence to development standards” is not common enough for safe, widespread testing.

The National Highway Transportation Safety Administration’s automated vehicle guidance has just 12 testing safety suggestions, and although the NHTSA encourages AV companies to submit self-assessments regarding these 12 elements, few do. Additionally, the AV guidance has no given metrics for autonomous driving system developers to understand whether or not they have effectively achieved all safety goals related to those 12 areas.

Although NTSB members are glad Uber is currently cooperating with the investigation after the incident’s findings were released late last year, they believe Uber has an overall “ineffective safety culture” that led to the fatal crash.

A First Look into Kodiak Robotics’ Self-Driving Longhaul Trucks

August 16, 2020 by Levinson and Stefani Leave a Comment

Kodiak Robotics, a self-driving truck company based out of San Francisco, has finally detailed its recent approach to the challenge of giving long-haul trucks self-driving technology after its 350th commercial delivery.

The company has released a 49-page safety self-assessment report demonstrating its commitment to public safety. The report gives a detailed explanation of how the startup operates and programs its autonomous vehicle technology.

“We believe it’s critical that we begin the process of explaining to the general public not just how we are safely testing our vehicles, but how we’re going to prove, mathematically and in plain English, that our vehicles are comprehensively safe even without a person behind the wheel,” said Kodiak co-founders Paz Eshel and Don Burnette.

This candid self-assessment aims to soothe any skepticism among the public and build trust and rapport with other motorists on the road.

“Given our progress, we think it’s time for Kodiak to start talking more about our unique, truck-focused technology,” said Burnette, who has previously worked for companies such as Otto, Uber, and Google.

Kodiak uses computer simulation and test track runs to examine and analyze its technology. The company was founded in April of 2018, had its first on-road test the following March, and made its very first delivery in July of 2019. Now, it has a fleet of 10 trucks hauling commercial freight throughout the state of Texas. Kodiak is also one of only a few trucking startups working to bring automation technology to long-haul trucking.

The startup’s technology often mimics how human drivers operate vehicles, such as within its ability to use lane markings as visual cues, rather than built-in map programming. This system within the “Kodiak Driver” allows for the vehicle to respond to and avoid unexpected obstacles on the road, like construction projects, even when the program’s maps don’t show them.

This ability is referred to as “perception over priors–i.e., that the Kodiak Driver trusts its eyes, not its memory, and we believe it represents a significant step forward for the AV industry,” said the company.

In its safety assessment, the Kodiak team shares its reasoning for focusing heavily on “structure highway driving” rather than city driving, as well as how its specifically-designed systems benefit heavy-duty trucks.

“We optimize our driving for safety, not comfort: paper towels don’t care if they get jostled a bit, whereas a rideshare passenger expects a smooth ride,” said the company.

In 2017, the National Highway Traffic Safety Administration released “Voluntary Guidance–Automated Driving Systems 2.0: A Vision for Safety,” which included a suggestion for any companies aligned with automated driving systems to provide a public assessment of their approaches to safety.

Since this recommendation, nearly two dozen companies have submitted these assessments, including other automated commercial vehicle competitors of Kodiak, like Waymo, TuSimple, and Starsky Robotics–although Starsky has since gone out of business.

“At Kodiak, we are strategic about every mile we drive,” said the company in its report. “We never drive our trucks for the sake of just logging more miles. Of course, this disciplined approach means we will probably never log as many test miles as some of our competitors. We see our lower mileage count not as a risk, but as a sign of our commitment to safety.”

Also present in Kodiak’s report was its explanation of its commercial-grade steering column “designed specifically for trucks,” which will eventually expand into dual-redundant electric motors that will allow the automated system to stay easily controlled if one motor fails. The system’s sensors also have “overlapping fields of view, so that every region around the truck is seen by multiple sensors,” which will allow the Kodiak Driver to always know the details of its surroundings, even if a sensor might fail. The AV technology will actuate the vehicle on “redundant, fault-tolerant computers,” that will run independently from the main computer. “These computers always know how to bring the truck to a safe stop, so that if the main computer should ever fail, they can safely achieve a Minimal Risk Condition,” the report stated.

Still, deploying a fleet of self-driving trucks without a safety driver on board is not likely to happen within the next year, Burnette explained.

“The fact is, once people get used to them, self-driving trucks will actually be pretty boring,” he said. “They’ll largely stay in the right lane, they’ll never weave in and out of traffic, and they’ll never speed.”

Supporters of automated vehicles within the commercial truck industry say these trucks will be much safer than human drivers, who often run red lights, text, fall asleep at the wheel, or otherwise become dangerously distracted while on the road.

“Our mission is to build the world’s most efficient, reliable, and respected freight carrier, using our autonomous technology,” explained Burnette. “Operating as a carrier will allow us to design our technology to meet shippers’ needs, and ultimately allow us to build a better product.”

NTSB Issues Safety Recommendations After Fatal Uber Automated Vehicle Crash

December 11, 2019 by Levinson and Stefani Leave a Comment

The idea of self-driving vehicles has been in the works for many years, and has recently come into full fruition, with multiple manufacturing companies jumping on the futuristic trend.

However, federal regulators are now being pressured by the National Transportation Safety Board to put in place a new review process for automated test vehicles after an Uber automated test vehicle hit and killed a pedestrian.

Last year, the NTSB released its preliminary report regarding its investigation of the fatal crash, which occurred in Tempe, Arizona between a modified 2017 Volvo XC90–which was occupied by a single vehicle operator but was running on its computer-controlled self-driving system–and a pedestrian in March 2018.

While the vehicle operator wasn’t hurt, the 49-year-old female pedestrian suffered fatal injuries.

According to the report, the pedestrian was wearing dark-colored clothing, didn’t look toward the vehicle until the moment before the impact took place, and crossed the road in an area without direct lighting. The pedestrian was also pushing a bicycle that did not have side reflectors, although it did have a front and rear reflectors that were positioned perpendicular to the oncoming vehicle’s path. She also didn’t use the nearby crosswalk, but rather entered the road from a brick median. Additionally, the pedestrian’s post-accident toxicology test showed both methamphetamine and marijuana in her system.

As for the test vehicle, the report said Uber had equipped it with an in-development self-driving system that was comprised of forward- and side-facing cameras, radars, navigation sensors, Light Detection and Ranging, and a data storage unit. It was also factory-equipped with Volvo Cars’ driver assistance functions, such as collision avoidance with automatic emergency braking, driver alertness detection, and road sign information. These functions are only disabled when the test vehicle is in computer control mode.

The data from the self-driving system showed that the car’s vehicle operator intervened by grabbing the steering wheel less than a second before the impact, which occurred at 39 mph. The operator also began braking less than a second after the impact.

At 1.3 seconds before the impact, the self-driving system did determine emergency braking was needed, but these maneuvers are not enabled when the vehicle is being computer-controlled. The vehicle operator is expected to take action at the point, as to reduce the possibility for erratic vehicle movement. The system also does not alert the operator of the need for emergency braking.

During this month’s board meeting, which was held in order to determine the probable cause of the crash, the NTSB said an Uber division’s “inadequate safety culture” is what allowed the fatal collision to take place.

The NTSB found that the immediate cause of the collision was the Uber ATG operator’s failure to monitor the road and the automated driving system closely enough–which it says was due to the her being distracted during the trip by her cell phone.

The NTSB also says Uber ATG held inadequate safety and risk assessment procedures, had a lack of adequate mechanisms for addressing vehicle operators’ automation complacency, and gave an overall lack of oversight of its vehicle operators in general.

Here are the investigation’s findings:

-The automated driving system was able to detect the pedestrian a full 5.6 seconds before impact. The system did continue to track the pedestrian up until the crash, but was never able to accurately determine what the object crossing the road was, or what its path would most likely be.

-If the vehicle operator had been paying close attention, she would likely have had enough time to effectively react to the pedestrian and either mitigate the impact or avoid the crash completely.

-Uber ATG managers rarely actively monitored the behavior of their vehicle operators, although they had the opportunity to do so. This oversight was made worse by Uber’s decision not to include a second operator in the vehicle during this testing.

-Uber ATG added a safety management system, among other updates, to address the present deficiencies.

The NTSB made six total recommendations to the National Highway Traffic Safety Administration, the American Association of Motor Vehicle Administrators, Uber ATG, and the state of Arizona–including that NHTSA requires developmental automated driving system test operators to submit safety self-assessment plans before they can begin operating on public roads. NHTSA will have to review these plans thoroughly to make sure all necessary safety precautions and standards are met.

Self-Driving Trucks to be Tested on Virginia Roads

September 23, 2019 by Levinson and Stefani Leave a Comment

VIRGINIA – In a partnership with Torc Robotics, Daimler Trucks will now be testing highly automated, heavy-duty trucks across highways in Virginia.

According to a September 9th press release, the two companies said they are currently actively developing and testing these trucks with SAE Level 4 intent technology on public routes in southwest Virginia, where Torc Robotics headquarters are located.

After months of testing and safety validation on closed-loop tracks, these public test runs of self-driving trucks will require an overseeing engineer and a highly-trained safety driver–who must hold commercial driver’s licenses and be specially trained in both vehicle dynamics and automated systems.

“We understand the needs of the industry,” Martin Daum, Member of the Board of Management of Daimler AG, said in the release. “Bringing Level 4 trucks to the public roads is a major step toward our goal to deliver reliable and safe trucks for the benefits of our customers, our economy and society.”

This announcement comes just months after Daimler agreed to acquire a majority stake in Torc–deeming Torc Robotics an official part of the new Autonomous Technology Group of Daimler Trucks, and allowing for these innovations.

Now, the truck manufacturer is consolidating its activities regarding automated driving globally–including within Torc’s locations in Blacksburg, Virginia as well as in Portland, Oregon and Stuttgart, Germany.

The Level 4 automated driving capabilities will allow trucks to drive themselves without needing a driver to monitor or engage–in certain conditions.

Torc’s system for automated driving, called ‘Asimov,’ has been tested throughout urban and long-distance routes in all weather conditions–including, rain, snow and fog.

“Our whole team is thrilled to be working alongside our Daimler colleagues as we pursue the commercialization of Level 4 trucks to bring this technology to the market,” said Torc Robotics CEO Michael Fleming. “We strongly believe it can save lives.”

The goal for these automated trucks is to eventually be able to haul freight on open highways between hubs.

The Level 4 test trucks are Freightliner Cascadia models complete with Torc’s self-driving capabilities, lidar sensors, cameras, and radar. During the tests, they will also pull weighted trailers to simulate carrying a load.

Daimler Trucks North America (DTNA) plans to evolve automated driving technology and vehicle integration for heavy-duty trucks. It is currently working on a truck chassis for automated driving, “particularly the redundancy of systems needed to provide reliability and safety,” the release stated.

DTNA will also be building an infrastructure to support Level 4 testing–which will include a control center and logistic hubs, located along high-density freight corridors where customers can operate near interstates and highways.

While Daimler continues with Level 4 automated driving, it will continue the process of integrating Level 2 technology, which aims to automate both lateral and longitudinal control while still requiring driver engagement.

DTNA will also continue offering active steering capabilities through its Detroit Assurance safety tech and Active Lane Assist by including automatic lane-centering and lane-departure protection features. This safety tech is projected to become more widely available early next year.

“As we pair Daimler’s expertise in building safe and reliable trucks with Torc’s genius in engineering Level 4 vehicles, we have no doubt we will do great things in the future,” said Daimler president and CEO Roger Nielsen. “We look forward to writing history together. The U.S. highways are the perfect place to develop automated technology.”

With the number of trucks on the road consistently growing but fewer people entering the profession, truck-driving continues to be stressful and often monotonous. “There is a long-term need to rethink,” Daimler says on its website. “Firstly, in order to increase the safety of all road users through the use of intelligent technologies. Secondly, in order to take pressure off the drivers and make their job more attractive by assigning them other tasks. Last but not least, the total cost of ownership (TCO) plays an important role in the transport industry.” Daimler claims the high degree of automation further lowers the TCO.

These public road tests are a significant step in Daimler Trucks’ 570 million dollar investment to bring highly automated trucks to global roads within a decade. Daimler says in this effort, it will be following its three most important beliefs: that “safety has absolute priority; everything must be 100% reliable,” that it must continue to develop “products together with its customers,” and that “a clear and legal regulatory framework for operation and liability is needed.”