Today's Top News: GM Plays It Safe with Cadillac Autopilot

GM Plays It Safe with Cadillac Autopilot

General Motors (GM) seems to have gone to great lengths to avoid lawsuits as it launches its first hands-off driving system in its soon-to-be-launched Cadillac CT6.

As GM makes the Cadillac CT6 available to a limited number of reviewers ahead of its launch this fall, early tests reveal the GM's Level 2 capabilities fall short of the range of driverless features Tesla models offer at comparable premium segment prices.

Cadillac's Super Cruise, as its name implies, serves more as an advanced cruise control than as a self-drive system. It will let the driver occasionally take their hands off the steering wheel during what GM says are "limited-access driving conditions" on highways only.

Super Cruise's main feature is it offers automatic lane-centering as well as self-braking and acceleration on highways. After an icon near the speedometer indicates the autopilot mode is available, the driver activates self-driving by pressing a button on the steering wheel. A green light on top of the steering wheel indicates the self-drive system is ready to takeover.

GM's first commercially available hands-off feature is limited in comparison to Tesla models. The Tesla Model S, X, and 3's self-drive mode works on highways, small roads, and at very low speeds on some city streets. They can also self-drive themselves in and out of garages and perpendicular park.

Meanwhile, GM offers a lot of advanced driver-awareness technology to make sure you keep your eyes on the road. In other words, GM has likely gone to great lengths to avoid someone suing it for negligence and other legal claims. This leaves the driver very little freedom to take risks by abusing the driverless features and then later suing GM, alleging it had not taken "reasonable" measures preventing drivers from misusing the system.

If the infrared rays detect the driver's eyes are not positioned on the road after seven to 15 seconds, a green light begins flashing, followed by a flashing red light, seat vibrations, and then different audible warnings until the car comes to a stop after activating the emergency warning lights.

Super Cruise's self-drive mode primarily relies on LiDAR sensors (The Tesla S, X, and 3 models produced since October have none), cameras, and radar sensors. For mapping, Super Cruise reportedly relies on data from GeoDigital, in which GM is an investor, which offers over 160,000 miles of highway data in the US and Canada.

We have not tested Super Cruise, so we have yet to judge how smoothly it rounds highway curves, maintains the Cadillac's position in the center lane, how quickly the self-drive system can take over after making a lane change, and other important self-driving performance benchmarks. However, we can say GM has not taken huge risks by offering advanced driverless features that are high on the "wow" factor.

Another Signage Hack Shown to Trick Driverless Cars

A group of researchers have published a paper showing how altering signs can be used to fool neural networks, representing a potentially serious safety threat. In one example, the researchers from the University of Washington, the University of Michigan Ann Arbor, Stony Brook University, and the University of California Berkeley showed how laser printing the words "love"and "hate" on white paper and then pasting the words on a stop sign caused a self-driving car's artificial intelligence (AI) system to confuse the signage for a speed limit sign. Meanwhile, another team of researchers recently showed how signs could be spoofed in more subtle ways that are invisible to the naked eye that also caused a neural network system to misread them.

Elsewhere in the News

Audi's A8 and Level 3 driving may not be legal before the end of the decade. The state of Ohio plans to build a 35-mile driverless highway. The UK creates tougher cyber security laws.

Cover photo via GM

1 Comment

My name is Michael DeKort. I am a former Lockheed engineer and DoD/DHS post 9/11 whistleblower.

Anyone who has an L2+/L3 vehicle is not playing it safe. It is just the opposite. It is dangerous, reckless and creates a massive false sense of security.

NASA and the aerospace industry, Volvo, Ford, Waymo and even Chris Urmson recently stated L2+/L3 should be skipped because it is dangerous. The data clearly shows it is not possible to keep people from being distracted. Ford found their professional drivers fall asleep and they couldn't stop it. That distraction requires 7 -24 seconds to over come. Even 7 seconds can be an eternity in a vehicle. The result is people making the wrong decisions or overreacting. Several airline tragedies have occurred because of it. What this also means by extension is that public shadow driving should not be allowed for levels 4 or 5 either. Aerospace level simulation should be used instead for AI, engineering and testing. A recent University of Michigan study showed even Grand Theft Auto, not nearly as capable as aerospace level simulation, could be used. What these companies and the press should do their due diligence and look in to this.

(Beyond the safety issues L4/L5 is not possible using public shadow driving because far too many miles need to be driven at too much of an expense. One TRILLION miles according to RAND and Toyota. At a very conservative cost of over $300B).

I am trying to resolve this situation by forming a nonprofit international Autonomous Vehicle and Mobility Simulation Association and Trade Study/Exhibit.

For more detail and data references please see my LinkedIn article

Autonomous Levels 4 and 5 will never be reached without Simulation vs Public Shadow Driving for AI
https://www.linkedin.com/pulse/autonomous-levels-4-5-never-reached-without-michael-dekort

Share Your Thoughts

  • Hot
  • Latest