Tesla Bot is exactly what we expected, for better or worse ? let’s find out
With its latest humanoid robot prototype, Tesla falls short of demonstrating anything particularly amazing.
Elon Musk unveiled a proposal for “Tesla Bot,” an electromechanically actuated, self-driving bipedal “general purpose” humanoid robot, at the conclusion of Tesla’s 2021 AI Day in August of last year.
Musk predicted that a Tesla Bot prototype, also known as “Optimus,” would be finished within the upcoming year. After much anticipation, a Tesla Bot prototype was really unveiled last night at Tesla’s 2022 AI Day. The hype was just that—hype, it turns out.
The humanoid robot that Musk briefly displayed on stage has absolutely nothing wrong with it, but it also has nothing particularly right.
We anticipated more from Tesla, if not necessarily hoped for it. And while the robot isn’t exactly a letdown, there isn’t much evidence to suggest that it will revolutionise robotics in the same manner that SpaceX revolutionised rockets or Tesla revolutionised electric vehicles.
We’re only going to concentrate on the parts concerning Tesla Bot/Optimus that are the most intriguing, but you can watch the complete three-plus hour live stream archived on YouTube here (which also covers vehicle stuff and other things).
Musk’s attempt to establish realistic expectations for this robot (or Tesla’s robotics programme in general) comes far too late. Most roboticists are aware that using humans to gauge humanoid robot performance can only lead to disappointment. While it is true that “compared to not having a robot at all, our robot will be quite amazing,” trying to preserve it at the last minute won’t help.
One of the developers made it clear just before the robot was brought onto the stage that this would be the first time it was walking independently and without support. If that’s the case, it’s crazy because there’s no reason to wait until now to give that a shot.
Many of the humanoid platforms in the 2015 DARPA Robotics Challenge reminded me of Tesla Bot in terms of appearance. In terms of form factor, I suppose there are only so many things you can do with a mostly naked electromechanical humanoid, but Tesla’s design doesn’t seem all that novel or forward-thinking at first.
The robot’s movement isn’t quite up to DRC standards because it appears as though it would struggle with any form of inadvertent touch or even a small amount of uneven flooring (and Musk suggested as much).
The robot didn’t do anything on stage. It managed to walk, although not really dynamically. We don’t know how much of its “moves” were planned, therefore we can’t say how much of the robot can balance by itself. Although I’m glad it didn’t, I wouldn’t have been shocked or passed judgement on it too harshly if it had.
Musk displayed several video footage of the prototype robot performing different tasks after the brief live demonstration (starting at 19:30 in the live stream). These videos showed the robot grabbing a watering can and walking while carrying a package of unknown weight and setting it on a table. The watering can was quite impressive because it seemed difficult to hold onto that small handle.
We don’t know how this was actually done, whether it was autonomous or not, or how many tries it took to get it right, despite the additional video from the robot’s sensors.
Another footage shows a robot picking something up and trying to put it in a bin, but the video ends before the attempt is accomplished. This leads me to believe that we are witnessing intentionally selected best-case performance circumstances.