Autonomous Vehicles: The Hubris of Artificial Intelligence
Autonomous Vehicles (AV) or a driverless vehicle, is a vehicle capable of sensing its environment and operating without human involvement. An autonomous vehicle utilizes a fully automated driving system in order to allow the vehicle to respond to external conditions that a human driver would manage. Autonomous cars rely on sensors, actuators, complex algorithms, machine learning systems, and powerful processors to execute software.
AVs have been touted as being safer, reducing accidents and deaths dramatically, yet it is unclear, beyond blue sky estimates just how much safer they really are. The promise may seem reasonable, but the deliverables fall far short.
While Tesla originally said that they expected their cars to be completely autonomous by the year 2017 the whole effort proved to be far more challenging than first expected. In 2022 In the US, there are only some 1,400 self-driving cars on the roads. Fully autonomous (Level 5) cars are undergoing testing in several pockets of the world, but none are yet available to the general public. We’re still years away from that and perhaps will not attain it. The technology systems necessary are very expensive. They may make the final cost of an AV prohibitively expensive for general public consumption.
The AV industry has operationally defined 5 levels:
Level 1 the vehicle’s ADAS (advanced driver assistance system) has the ability to support the driver with a single automated system (steering or accelerating and braking). Cruise control was an early example.
Level 2 has partial automation where the ADAS can oversee steering and accelerating and braking in some conditions, although the human driver is required to continue paying complete attention to the driving environment throughout the journey and can take control at any time. Multiple driver safety features that are now standard in many new vehicles including: lane departure warning system, parking assistance, automated breaking etc. each apply to one system with its own specialized function.
Level 3 has conditional automation where the ADS (advanced driving system) can perform most parts of the driving task in some conditions, but the human driver is required to be able to regain control. In the remaining conditions, the human driver executes the necessary tasks. There are a couple of manufacturers with this level of AV for sale to the public.
Level 4 has high automation where the vehicle’s ADS is able to perform all driving tasks independently in certain conditions in which human attention is not required. Human override is possible.
Level 5 is full automation where the vehicle’s ADS is able to perform all tasks in all conditions, and no driving assistance is required from the human driver. This full automation will be enabled by the application of 5G technology, which will allow vehicles to communicate not just with one another, but also with traffic lights, signage and even the roads themselves. No human interaction or attention is needed.
How do they work?
Autonomous cars create and maintain a map of their surroundings based on a variety of sensors situated in different parts of the vehicle. Radar sensors monitor the position of nearby vehicles. Video cameras detect traffic lights, read road signs, track other vehicles, and look for pedestrians. Lidar (light detection and ranging) sensors bounce pulses of light off the car’s surroundings to measure distances, detect road edges, and identify lane markings. Ultrasonic sensors in the wheels detect curbs and other vehicles when parking.
Sophisticated software then processes all this sensory input, plots a path, and sends instructions to the car’s actuators, which control acceleration, braking, and steering. Hard-coded rules, obstacle avoidance algorithms, predictive modeling, and object recognition help the software follow traffic rules and navigate obstacles.
- Debilitating problems not yet solved and which may not be solved in the foreseeable future of your organizations’ horizon.
- Much of the individual technologies systems are exceedingly expensive.
- Multiple overlapping systems for every function may be too complex and unaffordable for a passenger car.
- Multiple autonomous cars were to drive on the same road potentially can interfere with each other and other radar or electronic signals.
- Weather conditions (fog, snow, ice, dust, mud) can cover sensors rendering them useless.
- Human knowledge - drivers see a very different picture than sensors including verbal commands, language, eye contact, body language intentionality, understanding intentionality of other drivers.