The Autotech Council, based in Silicon Valley, is a community of approximately 100 companies, including major car manufacturers, who are committing to partnering with young companies and new technologies to improve their products and services, described Liz Kerton, the council’s executive director. Four start-ups presented their ideas during the KPMG forum at the Detroit auto show.
“A major safety problem on our roadways desperately needs the attention of the automotive industry,” according to Jeremy Agulnek, Vice President of Connected Car, at the start-up. “HAAS Alert is all about making our roadways safer for everyone.”
First responders and civilian motorists in the United States are involved in nearly 100,000 collisions with each other every year. In fact, first responders are at a greater risk of death and injury while traveling to an incident scene than at the scene itself. Construction and other workers on the roadway are also at risk, especially as distracted driving has increased with smartphone usage.
HAAS Alert leverages cellular networks to send data into what HAAS calls a “Safety Cloud” when a police, fire, or other fleet vehicle activates warning or emergency lights, Agulnek explained. HAAS then intelligently processes that data and sends an alert to drivers in proximity to the emergency or other vehicle.
The technology is sophisticated enough to not simply provide a radius-based alert, but rather, it can analyze the trajectories both the motorists and an emergency vehicles are taking, alerting only those drivers whose paths may intersect within 30 seconds.
HAAS Alert is live today for the 25 million users of the Waze GPS and traffic application, “but getting into Waze and other mobile apps is literally just the starting point for us,” Agulnek said. HAAS Alert seeks to incorporate its data into the vehicles themselves, where the alert can be provided any number of ways to the driver, such as displayed on the navigation screen and through audible notifications.
This integration is important as vehicles become more autonomous, he added. “By delivering an alert 10 to 15 seconds ahead of that possible collision point, the autonomous vehicle system can make a safer and smarter driving decision, and even reroute itself automatically around a hazardous situation.”
Beyond helping drivers, components of smart city infrastructure such as traffic signals and streetlights also could benefit from HAAS Alert’s real-time data and analysis, Agulnek said.
“Our mission is to improve the safety and efficiency of autonomous mobility systems by understanding and predicting human behavior,” Humanising Autonomy cofounder and CEO Maya Pindeus explained. “It’s really about making autonomy safe for people.”
As autonomous vehicle (AV) technology develops, it’s important to consider that humans are good evaluating situations and reacting in seconds, while machines “identify obstacles, but they fail to identify all the subtleties of human behaviors,” she said. “This is exactly what we are tackling with Humanising Autonomy. We need a deep understanding of human behavior in order to be able to deploy autonomous vehicles in cities like London, New York, Detroit, Tokyo, and many more.”
Humanising Autonomy detects the full range of human behaviors using visual sensor footage, and then it applies deep learning to predict in real time what the pedestrian on foot, bike, or other mode of transportation around the motorized vehicle will likely do.
Today, the company can identify and recognize more than 150 body language and other characteristics by fusing behavioral psychology and novel artificial intelligence methods.
The technology also can be used globally because it considers that citizens around the world behave differently, Pindeus said. “It’s really important to take this contextual and cultural aspect of human behavior into account. By building a modular prediction system, we’re able to tune our system to take the prediction from one city to the other with a very small amount of data.”
More than setting its sights on the AV market, “we're tackling a human safety market,” she said. The company’s two core market streams today are Level 2 AVs and up, and pedestrian safety technology for vehicle-to-everything infrastructure systems. Clients to date have included Airbus, to improve the safety of ground vehicles across their manufacturing and production sites; Kyocera, to improve crosswalks around central Tokyo; and Ann Arbor’s public transportation authority, to complement the bus driver and improve safety.
Michigan-based start-up Ushr has produced the most precise representation of the U.S. and Canadian interstate roadmap in existence, according to Brian Radloff, Vice President of Business Development and Sales.
In contrast to the competition, Ushr approaches data gathering from the perspective of a surveyor, who requires far greater geo-spatial accuracy, than a typical GPS map producer, who simply needs to represent the roadway from a visual perspective.
Ushr’s HD map is accurate within 15 centimeters, with probably 95 percent of the points on the map accurate within three to eight centimeters, Radloff said. By comparison, Ushr’s closest competitors advertise accuracy at around a meter. The map’s details also include lane markings, roadway edges, slope, and other details necessary for AV technology.
While much attention has been paid to the robo-taxi and shared mobility markets, Radloff said Ushr is focused on the autonomous personal mobility market.
The company launched its map in October 2017 in the Cadillac CT6 Super Cruise, a highway pilot product.
“We see the robo-taxi market as very interesting to us, but commercial deployments on a broad scale are probably still several years away,” he said. “This is a revenue path for us today, while we are still building out a map.”
Radloff added that robo-taxi vehicles are still in proof-of-concept phases, and once manufacturers prepare for mass scale deployment they will consider commercial solutions to help drive cost reduction. “We’ll be there with a map for the urban environments when they’re ready to deploy across multiple environments.” Ushr’s next phase involves mapping all state and federal highways, and then secondary and tertiary roads, by 2023.
Ushr also closed a first round of funding in November 2017 for $10 million, doubled its staff within the last year, and raised its profile. It also earned a number of awards, including the GM Supplier Innovation Award, the first start-up in history to do so.
Karamba Security believes it has found a better way to prevent hackers from compromising vehicle safety, according to Bryan Short, Vice President - North America.
WiFi, 5G, Bluetooth, DSRC, etc., are all gateways for hackers to get into the vehicle, and most companies have taken an information technology (IT) and data center approach with traditional intrusion detection, he said. But that’s like putting up a 20-foot fence with surveillance cameras between your house and the hackers, while the house is still vulnerable.
“My valuables, all my personal and professional data, might be in that house. I’m locking the doors and the windows—and that’s what we decided to do. We’re going to protect the car at the end point, and we do it on the individual ECUs.”
Karamba helps ensure the ECU or engine control unit maintains factory settings until the manufacturer makes a change or updates, spotting and shutting down attempts by malware to change the code, Short said. While IT can take days or even weeks trying to remediate a cyber attack after the fact, “we detect and then prevent the attacks from actually occurring.”
The patented process adds less than 5 percent to overhead costs, depending on the size of the code. As importantly, no change to the existing architecture is required. ISO 26262 standards for release this year include ASIL D requirements for certain autonomous and connected vehicle components, potentially driving demand for Karamba’s ISO- and ASIL D-compliant technology, Short added.
The company also has introduced additional security technologies beyond its initial ECU protection capabilities, including SafeCAN network protection allowing the hundreds of individual ECUs within a given car to communicate effectively with encryption, as well as its ThreatHive command center used to track, anticipate, and test systems against hacking attempts.
To access each of these presentations, please click here.