In the 1980s, pop-culture icons like the Terminator and Knight Rider propagated the concept of fully autonomous robots and cars. In the decades since, we’ve accepted the concept of autonomous vehicles as a coming reality, but they’ve still been limited to Sci-Fi.
But today, as sensors, computing, and actuator technologies rapidly advance, generalized machines have already transformed their efficiency and performance via automation and optimization. Autonomous vehicles (AV), autonomous robots, and autonomous cars all share the same common denominator technologies, such as machine learning algorithms, high-tech sensors, and functional requirements of autonomy with minimalized human interaction.
Given that the technologies between these devices are synonymous, this article takes a look at the history of self-driving cars and autonomous robots, and how that will guide the coming age.
The First Self-Driving Robot
Modern autonomous robots and vehicles operate using machine learning algorithms such as neural networks, many of which are based on the actual organization of the human brain. It’s no surprise that one of the first autonomous robots conceived was created by a neurobiologist.
William Grey Walter, an American/British scientist with a background in neurophysiology and cybernetics, developed the first robots that were designed to ‘think’ the way humans do. The robots, Elmer and Elsie, generally reflected the shape and size of two turtles and were designed to respond in specific ways to stimulus from light.
Elmer and Elsie were built shortly after World War II using surplus military parts and alarm clocks, yet their impact on robotics would go on to influence the next generation of autonomous robots and vehicles. These robots demonstrated one of the simplest (and very first) forms of autonomous navigation.
The First Autonomous Industrial Robot
The first industrial robot, Unimate, was introduced by inventor George Devol in 1954, and first put to work on a General Motors assembly line in 1961. Unimate was responsible for automating die-casting cooling. Newly forged (and very hot) door handles were dropped into a vat of cooling liquid, which eliminated the need for a human to be responsible for this very dangerous and potentially harmful task.
Unimate’s systematic tasks were stored in a drum memory device. It featured a computer-like box that helped it complete the necessary functions of its positions. Unimate helped the industrial and manufacturing sectors understand the value of autonomous robots, forging the foundation for nearly all task-oriented autonomous robots that you may see on assembly lines around the world today.
The First General Purpose Autonomous Robot
Shakey, the first real-world culmination of a personal assistant robot able to help with monotonous tasks, was built between 1966 to 1972 at the Artificial Intelligence Center of Stanford Research Institute. This invention was the first autonomous object that used logical reason to yield physical action. Shakey was a technical combination of computer vision, natural language procession, actuators, and sensors. It was able to take basic commands from a human and break those commands down into simple actions, and then break those actions down into specific action sequences via on-board systems.
While this may seem novel, this information breakdown is still seen in modern autonomous vehicles. For example, Tesla automobiles have a ‘summon’ feature in which a user can summon their Tesla to autonomously navigate to the summoner in a parking lot. The Tesla is able to break that command into a series of specific maneuvers, which all consist of specific system control commands.
The First Self-Driving Car
Navlab5 is the minivan in the foreground.
The first self-driving car was successfully piloted from Pittsburgh to San Diego in 1995, although it was not a standard production vehicle. The Navlab5, which was the fifth of ten cars that were retrofitted by The Robotics Institute at Carnegie Mellon, was an autonomous vehicle that was able to steer itself. Of the nine other Navlab vehicles, some were semi-autonomous while others were fully autonomous in limited scenarios.
Navlab5 is known for steering itself from Pittsburg to San Diego without any added human input, making it the first autonomous car. While the actual driver of the vehicle was responsible for acceleration and braking, Navlab5 was able to utilize cameras and a variety of other sensors to navigate over 2000 miles without human intervention, which is considered a monumental milestone in the history of autonomous vehicles and robotics.
First Vehicle with Intelligent Obstacle Avoidance
Perhaps considered both an autonomous robot and an autonomous vehicle, the Mars rover Sojourner was the first autonomous device that was able to intelligently sense, identify, classify, and navigate through specific object obstacles. Sojourner was developed by NASA in the early 1990s and landed on Mars on July 5th, 1997.
Given the 5 to 20 minute communication delay between Mars and Earth, the Mars rover necessitated navigational intelligence that would allow it to drive autonomously without human interaction. While the Mars rover was piloted by scientists back on earth, they would simply give the rover directive on where to move to next, instead of specific navigation paths or strategies.
Sojourner utilized a laser and cameras to identify and measure objects to better understand how to maneuver to the next waypoint. While object detection has advanced leagues beyond the Sojourner’s technology since 1997, the laser-based detection technology that was used in this Mars rover was the first of its kind and a foundational pillar for modern LIDAR technology and object detection methodologies.
The First Production Self-Driving Car
The claim to the first production self-driving car depends on the definition of autonomy. There has yet to be a production vehicle that is capable of Level 4 or 5 autonomy, but there are many cars on the road today that are capable of Level 3 self-driving. Cars that were first capable of Level 1 and Level 2 self-driving reached production by the late 1990s through the mid-2000s. For example, the 2006 Lexus LS 460 was the first production autonomous car that was capable of active lane-keeping assist, meaning it could fully steer the vehicle and correct it if it was moving out of the desired lane. Honda was able to achieve this feat in 2003, but only applied 80% of the torque needed to steer the vehicle. These early lane-keeping assist technologies relied on cameras mounted above the windshield to understand a vehicle’s position in a lane.
Modern Autonomous Robots & Vehicles
There are hundreds of companies working towards fully autonomous vehicles, but in many industries, fully autonomous robots are already fully functioning. Dark warehouses consist of fully autonomous robots and vehicles that operate entire warehouse facilities with little to no human intervention. Modern manufacturers in nearly every industry around the world rely on autonomous robots to perform specific tasks with a dexterity and efficiency unachievable by humans.
Meanwhile, many industries—including the commercial automotive world—are integrating varying degrees of autonomy into our everyday lives. Less than a century ago, Elmer and Elsie were simply following light around a room aimlessly. Now we have semi-autonomous and fully autonomous machines in every corner of the planet.