The rise of robotics has been outstanding in the past few years, and we can nod our heads to several vital aspects that have accelerated both adoption and development:
Open Source Community
The robotics community has a strong foundation in open source, likely due to higher industry moats and the difficulty of copycats.
AI/ML Frameworks & Software Frameworks
Significant progress in AI and ML algorithms has enabled robots to perform complex tasks more efficiently and with greater autonomy, and frameworks such as ROS (Robot Operating System) have improved efficiency and standardization.
Decrease of Component Cost
The cost of sensors, actuators, and other critical hardware components has decreased substantially, making robotics more accessible and cost-effective for a broader range of applications.
Since many other factors are increasing accessibility, it’s no surprise that we are starting to see massive VC investments into the space, especially after some high-profile acquisitions, such as Spyce by Sweetgreen, Bear Flag Robotics by John Deere, and Kiva Systems by Amazon. Companies are now trying to take the route of trying to be the last word in robotic automation.
One path to this goal is through creating humanoid robots. Many high-profile investments in space, such as Sanctuary AI, Agility Robotics, the Tesla Bot Optimus, Figure, and Boston Dynamics, have all had their shot at creating humanoid robots.
The question remains: are humanoid robots the last word in automation? Or are they another VC hype cycle where the product is defined before the problem once again?
Let’s dive in:
We’ve had robotics try to replace human motion profiles for a while now, with the first industrial robot-arm-looking thing being Unimate. Since then, robotic arms have been a popular way to manipulate an object and move it from point A to point B. But let me pose a question:
Let’s say you want to move a soda can from point A to point B. A common way an engineer would look at this problem was to reduce the object to a dot and observe how the dot is desired to move. Rather than moving the dot with a trapezoidal velocity motion profile with a robotic arm, why not just use a conveyor belt? It accomplishes the same goal of moving the dot!
This is a common problem in robotics: do you adapt the environment for automation? Or do you adapt automation to work with the current environment? This question poses a fundamental risk to the hypothesis of humanoid robots and, if true, cuts their projected TAM in half.
If the end goal is a fully automated warehouse, why not change the warehouse's layout and machines if humans are not in the equation?
Adapting automation to the environment
Robotic solutions where the environment is pre-set and unable to be changed call for robotic solutions that can adapt to the environment. Some good examples are Bear Flag Robotic’s adaptable robotics suite for tractors, Miso Robotics’ arm used to fry over a grill or fryer, or a Roomba mapping out a floor through trial and error and barriers. This is the path the humanoid robots are taking, focusing on being the last word on automation by enveloping the form factor that the world environment is adapted for: humans.
Adapting the environment to automation
When the environment is changed to support automation, you can create more flexibility in the types and motion profiles, optimizing for efficiency and specific problems. Some examples are robotic coffee machines and robotic kitchens (like Spyce). These concepts are trying to remove humans entirely in the end-to-end process. If full automation is the goal, investing in adapting to the environment should be a focus.
Humanoid robots will have a place, but the money invested & the product design doesn't make sense. It’s long understood that legs aren’t the most efficient type of movement, so why are we designing these robots with them?
While I haven’t seen a deck for a humanoid robotics company yet, I’m sure their market potential slide does not factor in the infrastructure change we will incur to adopt automation that will create less need for humanoid robots.
Thank you all, and see you in 2 weeks!
— The Prompter Team