How long can Optimus work between charges?
The most recent FUD, is that Optimus cannot operate for 8 hours, or even close to that. Let's take a look.
FUD Argument One: The battery pack cannot support the compute, let alone the actuators.
I am using a laptop to compose this post. It is using the most inefficient O/S known to mankind: Windows. Windows is not only inefficient, it manages a plethora of inefficient apps, known as bloatware, in the foreground and background. Despite this behavior, my laptop can operate 2 hours unplugged doing normal work, 6-8 hours in standby, and about 1 hour doing intensive computation and screen display. With a battery equivalent to a 2170. Optimus has 140 of these batteries, so it could run my laptop for about a week.
My phone is also a marvel of computing. It cannot only take calls, but stream videos. I do this daily on 1-2 hours walks with plenty of battery left when I return. A battery far smaller than a 2170.
The Optimus computer is not a laptop, and does not need to manage bloatware. It does not need to stream YouTube videos or make Zoom calls. It does not even have a screen. It is a bespoke O/S using an inference engine. Inference is far more compute and energy efficient than heuristic code written for a traditional CPU. The FSD chip uses less than 75 W. Even using the laptop analogy, eight 2170's at full compute gives 8 hours of operation. We still have 132 left in reserve.
In addition, I use my laptop battery when I am mobile, but plugged-in when I am stationary. Nothing prevents the same for Optimus.
Optimus expects a maximum power usage of 500 W. With a 2.3 kWh pack, that is a little over 4 hours at full power. But the bot will not be running at the maximum for 100% of the time. At an average usage of 250 W it is feasible to run for 8 hours.
Optimus does not have a swappable battery pack. It is the swappable battery pack. For every 8 Optimi, one can be charging, then tap the shoulder of the Optimus in the lowest state of charge and take over that task. Sort of a Pony Express of manufacturing. The price of Optimus will make this affordable.
FUD Argument Two: The actuators are power hungry.
Remember the Optimus Actuator lifting a Half Ton Grand Piano (video below)? How much power do you think that requires?
A half ton, is 500 kg. From the video the piano appears to move 75 mm in about 2 seconds. Power = Weight x Velocity.
Power = 500 x 9.8 x 0.075/2 = 183.75 W
Less that 200 W is all that is needed to lift that grand piano. For context, your blow-dryer uses 1500 W. At 200 W, there is enough energy in the Optimus battery pack to lift the grand piano for 12 hours. Or lift it to a height of 276 meters.
The 2.3 kWh battery pack is about 1/30 of a Model Y battery pack, and has enough juice to propel a Model Y for 10 miles at highway speeds.
Optimus will not be lifting grand pianos all day long nor walking 10 miles, but doing lighter tasks with less walking. Suffice it to say, it has the energy to do a shift.
Tesla needed to design bespoke actuators, because the market does not provide actuators optimized for humanoid robots. Yes, Bumblebee proved one could use off-the-shelf actuators as a development platform, but to get required performance and scalability one needs custom actuators. This is the reason for the low power usage. Figure, and likely others, has also reached the same conclusion regarding bespoke actuators.
Plus, the argument that the bot will be expensive because of the number of actuators is also a canard. The Model 3 has a similar number of actuators as Optimus, yet can be produced for under $35K. Tesla has a new drivetrain for the next Gen car which can be produced for $1000. Optimus actuators are far less complicated.
FUD Argument Three: Optimus is defying the Laws of Physics.
Mercedes tried that argument regarding the semi, and that didn't age well. But don't just take Tesla's word for it, listen to Agility, or Sanctuary, or Figure, or Apptronik, or Unitree, or Fourier or numerous other humanoid robot vendors. Even with smaller battery packs they expect 3-4 hours of operation.
The work day is a marathon, not a sprint. Work tasks are not high intensity workouts. In a properly designed work environment, humans workers may be tired, but not exhausted, at the day end. Same for the bots.
As I will outline in a coming post, there are other strategies that will allow humanoid bots to easily manage an 8 hour shift, and still be ready for the following shift.