Elon Musk's "Alien Dreadnought" factory could see a second when the Tesla Model Y is ready for the production ramp, thanks to the latest advances in data vision.
Before Tesla started the model 3 ramp, Elon Musk explained his vision for a hyper-automated factory which he described as "the machine that builds the machine." Appealing investors, Musk revealed that the project's code name was "Alien Dreadnought", a reference to the futuristic spacecraft commonly used by extraterrestrial people in sci-fi movies. Musk planned to use Alien Dreadnought for the Model 3 ramp, and he expected the automatic factory to be operational by 2018.
This timeline proved to be too optimistic, with the Tesla CEO later admitting he had been overly enthusiastic with the Model 3 production line automation. "People are undervalued," Musk admitted in an interview with CBS . Finally, there was a combination of traditional human work and non-Alien-Dreadnought type automation that became a breakthrough for the Model 3 ramp. Tesla has since gained access to model 3 production, with the company producing enough cars for the United States, as well as Europe and China (at least for selected regions) today.
According to Ryan Kottenstette, who serves as CEO of Cape Analytics, a group that performs computer vision research, the challenges Tesla faced with the model 3 ramp, were caused by the limitations of robot vision. Kottenstette notes that the robots Tesla used in the Model 3 line simply could not handle unexpected events in the production process, such as different orientations on nuts and bolts, or complicated maneuvering between the car frame.
Musk mentioned one of these challenges in the revenue number Q1 2018 when he described a machine he instantly called "Flufferbot", which had many problems separating glass fiber mats (or "fluff" as Musk described it) for the model 3 battery pack. Musk stated that Flufferbot eventually lost his job because it became apparent that scooping "fluff" is far easier when it is done with human hands.
Kottenstette claims that the data vision has evolved significantly over the past few years, with approaches such as Generative Adversarial Networks (GAN) and uncontrolled learning that expand the data vision envelope in applications, robustness and reliability . These advances can help solve some of the challenges facing Tesla under the Model 3 ramps. Along with Tesla's technology skills and enhanced by the company's rugged AI team, these new data vision approaches can be utilized to allow a new attempt on Elon Musk's "Alien Dreadnought" plant.
An opportunity is actually foreseen for Tesla, as the company is scheduled to uncover model Y SUV next week. Like model 3, model Y is a hugh volume, mass market vehicle, and Elon Musk expects the demand for the vehicle to go up to 1 million units per year. Unlike Model S, X and Model 3, Tesla is set to launch the first Model Y production on Gigafactory 1 in Nevada, meaning the company can design and set up SUV production lines from an empty slate. Thanks to model Y, Elon Musk's "Alien Dreadnought" factory may end up being closer than we think.