The current industry of driverless vehicle development has been unable to produce an independent working model.
Representatives with companies like Uber and Google have made claims that a driverless vehicle would be available and in operation on public streets before 2025.
However the the technology and it’s development thus far, have proven driverless vehicles face technical challenges that the worlds best engineers are still decades away from developing before they could ever be effectively tested on public streets.
Red Whittaker is the first major developer of driverless vehicles in the US. Having contributed to several projects with DARPA, the US Army and DOD he has worked with all of the primary systems that make up what a self guided land vehicle is in it’s current incarnation. During his time developing these vehicles it was determined what would be required of a driverless vehicle to truly be “driverless”.
For a driverless vehicle to fulfill it’s purpose and be economically viable for it’s cost and operations, it must fulfill specific functions:
1- Operate autonomously WITHOUT the need of a human driver or human live monitor from a remote location.
2- Operate independent of any failures with one or more sensory systems with redundant and/or alternative systems.
3- Operate safely and independently along side Human operated vehicles at all posted speeds.
4- Operate in all weather conditions.
Current driverless vehicle technology has not accomplished ANY of these goals
Recent failures of the technology have indicated clearly that it can not perform as designed or intended.
Though the accident was the fault of the Human truck driver the driverless vehicle failed to sound a horn in warning or back up in available space behind it to avoid the truck as it slowly backed into it.
Tesla’s ‘auto pilot’ functions have failed in two major incidents in the last two years.
THIS MONTH, the Phantom AI L2 system failed during a highway test. The human driver had to take over as the system could not properly identify or react to an object suddenly falling in the road SEVERAL car lengths ahead of the vehicle.
A recent MIT study of LIDAR, the primary system by which driverless vehicles are supposed to navigate and judge their surroundings showed significant issues with the technology’s ability to safely guide vehicles.
In each of these failures the obvious issues with driverless vehicles becomes clear.
1- The LIDAR systems do not effectively calculate distance or discern motion.
LIDAR is the laser radar scanning system that these vehicles use to judge distance and direction for objects around it. The numerous collision failures indicate that the trajectory of objects and other vehicles around the vehicle are not being properly logged or accounted for.
Reaction time of the vehicles navigation system to respond to input from Lidar
2- Bad weather significantly inturupts the vehicle’s ability to operate safely.
3- The vehicles can fail entirely if just one sensor system is not functioning properly or is tricked by outside stimuli, it does not understand.
4- The vehicles are not actually thinking or making proactive judgments about their course or method of travel.
Current technology for autonomous vehicles is still so poor that when they are in traffic with human operated vehicles, they themselves must be operated by humans to avoid accidents.
Not a single area that has introduced these vehicles to regular street traffic has seen them operate without a live human behind the wheel.
This is because autonamous vehicles lack the one thing that would ever allow them to operate safely free of direct human oversight.
That being, an artificial intelligence which can operate a driverless vehicle BETTER than a human being based on auditory and video input ALONE.
Because Lidar, radar and GPS navigation can give false readings and be confused so easily by outside stimuli a truly “driverless” vehilce would have to be able to navigate and judge navigation based on the same methods of sensory input Humans use.
It would also have to proactively THINK and judge navigation BETTER than a human.
Current automated vehicles simply do not operate this way. Nor would they ever be able to without an artificial intelligence that is compact and thanks better than a human being with the same ability to problem solve and act quickly in a manner that is proactive as apposed to only reactive. This system would also have to be paired with an identical system acting in redundancy in case of a failure with the primary system.
This manner of artificial intelligence simply does not exist. Nor will it exist for some time.
If it did exist, one of the last things science would be using it for would be to navigate vehicles.
Such an AI would be the most expensive on the planet. It would be capable of activity solving complex math and engineering problems in a fraction of the time a human can and without any need for physical models or experimentation in order to formulate working models.
However such an AI in it’s first incarnations would definitely not be small enough to fit into a car.
At the dawn of the 21st century a new breed of snake oil peddling has taken root in silicon valley.
Concepts of science and engineering that seem plausible but their application in the real world is quite impossible and often times impractical.
Millions and in some cases BILLIONS of dollars have been invested into these ventures with no return on the actually technology being developed or actually working.
Since June of 2016, Ride Safe has issued a challenge to ANY and ALL developers of driverless technology to complete a series of just FOUR, no contact test challenges to driverless vehicles which would all be easily passed if the technology was viable.
So far NO ONE has accepted or responded to our requests to accept the challenge.