Tesla Makes a Leap from Autonomous Vehicles to Robots with “Tesla Bot” Reveal
Not a stranger to taking on a challenge, Elon Musk reveals "Tesla Bot" at 2021's Tesla AI Day. While it seems to be an ambitious plan to build a humanoid robot, how feasible is it for Tesla?
Elon Musk isn't new to having big ideas and even bigger ambitions. Keeping up with expectations, he announced his latest initiative of a revolutionary implication at the latest Tesla AI Day [video]; a humanoid robot servant dubbed the "Tesla Bot."
The concept design for a Tesla Bot. Screenshot used courtesy of Tesla [video]
The announcement has been a cause for both excitement and skepticism, and the real question on everybody's mind is: can Tesla do it?
Musk seems to think so, but his critics think he may be naïve and over-ambitious. This article will aim to cover Musk's plan for the Tesla Bot and see whether or not Tesla could be poised to pull it off.
Plans for Tesla Bot
Describing Tesla Bot, Musk laid out some pretty ambitious goals, namely goal for a humanoid, bipedal robot that can relieve humans of tedious tasks like grocery shopping or dangerous tasks such as manual labor.
To this end, Musk described his robot as 5'8" tall, 125lbs, and moves at a maximum speed of 5 mph. This robot is intended to have a carrying capacity of 45 lbs and a deadlift capacity of 150lbs.
While this sounds simple enough at an extremely high level, the question is if the tech is out there to make this from science fiction to reality.
From Autonomous Vehicles to Autonomous Robots
One thing that Tesla has in its favor is its expertise in AI and autonomous systems because, from this perspective, the task of self-driving and the task of creating a humanoid robot is not so different.
Broken down more simply, self-driving requires a car to view, map, interpret, locate itself, make decisions, and navigate its environment. For this, Tesla uses a purely camera-based system that feeds into a HydraNet: a multi-headed neural network that can recognize dozens of objects in the provided images and make decisions based on these outputs.
A major challenge in both self-driving and robotics is Simultaneous Localization And Mapping (SLAM). Image used courtesy of MathWorks
From a perception standpoint, the Tesla Bot could hope to leverage much of the same hardware and software as an autonomous vehicle.
The neural nets would need to be reconfigured for the robot's specific tasks and environment; however, the underlying network architecture, the camera systems, and the compute hardware, could likely stay the same.
This reconfiguration of technology could make the development of the Tesla Bot significantly simpler than starting from scratch and is one reason why some feel optimistic about this project.
Despite the technology's availability, there are still some odds stacked against this endeavor.
Despite the similarities, there are undoubted differences that make a Tesla Bot very difficult to achieve.
For starters, getting a car to drive is significantly easier than creating a bipedal robot that can walk, balance, navigate and perform tasks independently.
Driving a car requires an engine to turn a wheel and axle system, though not trivial, but something that's been long done and understood. On the other hand, building out a working bipedal robot requires the use of a variety of actuators, most of which are not found in a car.
This pursuit requires electrical, mechanical, and software engineers with specific fields of expertise, including biomechanics, locomotion, and robotics, all of which Tesla, as a company, generally lacks.
Components of an AI robot are not often found in a car. Image used courtesy of TechVidian
Currently, one of the most state-of-the-art bipedal robots is Boston Dynamics' Atlas. This project has been in existence for over a decade and has required the undiverted attention of the world's foremost experts.
Even given all of that, if you've seen the outtakes from Boston Dynamics' videos, you'd know that Atlas is far from everyday autonomy. Critics take this as proof that creating a Tesla Bot mostly from scratch is an almost unattainable goal, especially in a couple of years.
Pushing Impossible to Plausible
Between reusable rockets, electric vehicles, and brain-interfacing integrated circuits, Musk has pioneered multiple engineering feats that may have been called impossible at the start.
So, while it is recognizable that Tesla has many challenges ahead of it, it is also important to remember that Elon Musk is no stranger to navigating uncharted territory and coming out the other end with a revolutionary product.
While there's no saying for sure whether or not Tesla will be able to pull this off, with enough smart people, enough desire, and enough money, most things are possible, and Tesla should have no problem in those departments.
I would have liked more technical details and less rehashing of Elon Musk. After all, this is “All About Circuits” and not “Let’s Talk About CEOs.
Perhaps we’ll see this in a more in depth, technical, article?