Autonomous Vehicles: An Idea That Improved Turned Into Reality

The software company Synopsys defines an autonomous car as “a vehicle capable of sensing and operating without human involvement.” What does that even mean? How do you create an autonomous car from scratch? What are the capabilities and inabilities of a self-driving car? How safe are self-driving cars and are they a logistical new type of transportation?

Six Levels of Automation:

1. No Automation (Level 0): Driver handles all human tasks, including steering, braking, managing acceleration, etc.
2. Driver Assistance (Level 1): Driver still handles most human tasks, but the vehicle contains a single automated single (Example: monitoring speed through cruise control).
3. Partial Automation (Level 2): The vehicle contains more autonomous features including ADAS (Advanced Driver-Helps) and systems for steering and acceleration. However, the driver can still take control.
4. Conditional Automation (Level 3): The vehicle controls most tasks and has environmental detection capabilities. However, like in Level 2, there is a human override.
5. High Automation (Level 4): Vehicle controls essentially every task in every circumstance imaginable. The vehicle creates an invisible boundary or geofence showing the vehicle if they can or cannot drive in that zone.
6. Full Automation (Level 5): The vehicle performs all human tasks under any condition. No human interaction is necessary.

Leonardo da Vinci

All these questions stem from the first self semi autonomous vehicle created by Leonardo da Vinci in the 1500s. Da Vinci’s designed the cart with the ability to move without being pushed or pulled. Springs yielded the cart’s power under high tension. The brakes and steering features had some mind-blowing capabilities as da Vinci built a steering system where the cart can either go straight or at a pre-set angle.

Designs of da Vinci’s Self-Propelled Cart from the Codex Atlanticus

Da Vinci’s cart was so advanced for its time scholars could not figure out if the cart had worked until the late 20th century and not until 2004 when the Institute of Museum of the History of Science in Florence, Italy, created the first working replica cart.

Whitehead Torpedo

The next ground-breaking invention in the history of autonomous vehicles was over three centuries later in 1868 when Robert Whitehead created a self-propelling torpedo in Croatia called the Whitehead Torpedo. This cigar-shaped torpedo could travel 200 feet underwater at 6 1/2 knots (7.5 mph) and maintain depth for a long period because of the pressurization dubbed “The Secret.” This torpedo became a necessity for countries with navies across Europe. Austria first ordered these torpedoes and other European countries including Great Britain, Russia, France, Greece, and Italy. Robert Whitehead created more self-propelled torpedoes later in his career, including MK2, MK3, and MK4.

World War II

Like in other fields, World War II sped up technological advancements in autonomous vehicles. There weren’t any significant improvements or new inventions during World War I, but starting in the 1930s and onwards there were significant changes in the slowly growing industry. In 1933 Wiley Post was the first person to to fly solo across the globe. His Sperry Gyroscope, more commonly known as “Mechanical Mike” was autopilot which incorporated two gyroscopes. One gyroscope was used to sense changes in pitch and another for heading changes.

Sperry Gyroscope

12 years later in 1945 Ralph Teetor became fed up with the rocking motions of his car, creating the first cruise control by counting the rotations of the vehicle’s shaft and to calculate its speed and then using a bi-directional screw-drive electric motor to adjust throttle and maintain pace. A decade later it was being used for commercial use first by Chrysler and then basically every major manufacturer.

The Calm Before The Storm

MQ-1 Predator

Beginning in the 1970s, the frequency of advancements in autonomous vehicles has increased. In 1977, Tsukuba Mechanical Engineering accomplished a completely autonomous car that reached speeds of 20 miles per hour by tracking white street markers. In 1988 Ernst Dickmanns created dynamic vision, adding cameras throughout his car which detected objects in the road.

Finally, in 1995, General Atomics created the MQ-1 Predator. The vehicle was an unmanned plane primarily used by the Air Force. This decked out airplane contains a Multi-Spectral Targeting System integrated with an infrared sensor, monochrome daylight TV camera, image-intensified TV camera, and laser designator and laser illuminator. The MQ-1 Predator communicates to a ground control station called the Predator Primary Satellite Link, which has a rated pilot to command the mission and take control of the aircraft along with crew members to operate the sensors and weapons.

Technological Eruption

Of course, starting in the 21st century there were some of the most significant technological advancements in the automotive industry. There were significant changes to cars, airplanes, and even drones. Starting in 2004, the U.S. Department of Defense's research arm, DARPA, held a challenge where you needed vehicles to self-navigate 150 miles of desert roadway. No car completed the route, but this challenge pushed autonomous vehicles forward.

In 2015, Tesla entered and created a semi-autonomous autopilot feature. This feature enabled the vehicle to have hands-free control on the freeway. This feature has twelve ultrasonic sensors monitoring traffic all around the vehicle, a camera in front of the windshield, and long-distance radar to detect what lies ahead. Tesla Autopilot in 2015 had four functions: Autosteer, Auto Lane Change, Autopark, and Side-Collision Warning.

Tesla Simulation of Autonomous Driving

The right major companies like Waymo, Cruise, and Tesla are all gaining authorization to test and make their autonomous vehicles for commercial use. The hub for autonomous vehicles is currently on the West Coast and specifically California, as most of these companies' headquarters are currently in Silicon Valley.

Major Autonomous Companies Progress Reports:

1. Waymo: Has been giving paid rides to civilians in autonomous vehicles with backup drivers since 2018. On October 8th, 2020, they released their fully driverless service to the public in Phoenix, Arizona.
2. Lyft: Lyft has given over 100,000 paid autonomous rides to back up drivers. However, because of the pandemic, Lyft halted testing of autonomous vehicles for about three months.
3. Cruise: Gives free-hailing services in San Francisco to employees. They are testing and testing autonomous vehicles at the end of 2020 in San Francisco after receiving a permit from the California Department of Motor Vehicles.
4. Voyage: Voyage just released their new generation of rob taxis in August 2020. This taxi’s three major components major components are the brain or Autonomous Point-to-Point System (Commander), Collison Mitigation System, and the Remote Operations Solution (Telessist).
5. Tesla: Tesla has just released its first Full Self-Driving Beta software to a limited group of owners. Elon Musk reports that in two weeks Tesla is releasing more widespread FSD Beta software.
6. Uber: Uber paused operations for autonomous vehicles after one of its vehicles fatally killed someone and got their testing license revoked in Arizona. Now in San Francisco, California, they resumed testing in March 2020.

Anatomy of Self-Driving Cars

Yes, both self-driving cars and regular vehicles have most of the same components, but self-driving cars have completely different sub purposes than the regular car. The five major components major components of an autonomous vehicle include computer vision. Sensor Fusion, Localization, Path Planning, and Control.

Computer Vision

Computer vision represents the eyesight of the car. It’s a branch of AI that allows computers to view the world around them. Computer Vision is essential for self-driving cars as one of the key human tasks of driving a car is to be aware of your surroundings. This could be as reacting to a stoplight turning red or another vehicle behind you changing lanes. Cameras execute this crucial task by taking pictures of your surroundings and sending them to the next component: Sensor Fusion!

Uber’s Autonomous Vehicle

Sensor Fusion

Sensor Fusion has two subcomponents for success. The first component is data from the camera and understanding and adjusting the vehicle based on the environment. The second component is incorporating data from sensors on the self-driving car which include Ultrasonic Sensors, Radar, Cameras, LiDAR, and GNSS. Ultrasonic sensors, radar, and LiDAR all measure the distance to an object and the GNSS gathers all the data together and le the vehicle process its surroundings. The difference between the three measuring sensors is that the LiDAR measures distances between vehicles and the radar measures the distance between metal objects.

Localization

Localization is essentially the car identifying its position in the world by using data from GPS, landmark positions, and data from the sensor fusion stage and using that data to create algorithms. The first prediction would be straight from the GPS data and then movements of the car, landmark positions, and a combination of other data pieces to create a more accurate position. This algorithm also works in rough weather using the Localizing Ground Penetrating Radar (LGPR).

A vehicle using its sensors to analyze its environmental surroundings!

Path Planning

The term Path Planning is pretty self-explanatory. It’s pathing your car’s path to the destination while using data from the sensors above and predicting if you need to change routes, change lanes, slow down, speed up, or stop. A general algorithm for Path Planning would be to first determine a rough long-range plan to your destination while planning a concrete short-range path based on your position and speed.

Control

The last component of the anatomy of self-driving cars is control! Control’s job is to remove the need for human help. It makes it possible for the vehicle to go on the path created and handle all the parts of driving a human normally would. Control is the big key versus regular cars as it’s more precise than a car operated by a human making them safer than classic cars.

Key Takeaways:

• Having a creative mindset is necessary to grow and to invent! Thinking outside the box will establish new thoughts and ideas you wouldn’t have thought of before.
• Developing something new won’t be immediate, you will need to go through a create, test, and re-engineering process multiple times to reach your end goal.
• Most people don’t understand the multiple layers behind objects they see and use every day. Having a deeper understanding of these common objects will enable you to think about other objects in a deeper sense or enable you to redesign those same objects.

Linkedin

Hi, I’m a 16-year-old outside of Boston, Massachusetts with a passion for Advanced Transportation! Some of my other hobbies include working as a First Responder, being an inactive member in Massachusetts DECA, and suffering from heart attacks each time the Boston Celtics play!