I love robots, at least in the movies. Pick a Sci-Fi movie, and I can probably rattle off the name of a whole bunch of nifty robots. Robots that help, robots that work, robots that talk and advise, tell jokes, heal, and of course, robots that torture and KILL.
I have my favorites for all of these. But that’s the movies. I mean, have you seen the videos from those Boston Dynamics demos? Who the heck would want one of those things running around the house, interacting with your children? My colleague, David Gewirtz, thinks it’s a bad idea.
For most of us, a full-fledged autonomous robot is simply just a futuristic dream. Consumer robots are relegated to the most mundane tasks, such as vacuuming up dust or detritus in the pool. My Shark vacuum cleaner is as dumb as a rock and totally harmless.
Elon Musk, the founder of SpaceX and CEO of Tesla, wants to change that. Last week, we saw him get up on stage and talk about features and specifications for a new humanoid robot, “Tesla Bot,” that his company will supposedly release within the next several years.
Musk claims that Tesla will have a prototype ready in one year; it will weigh approximately 125 pounds, measure about 5 foot 8 inches tall, have a screen for a face, and be designed to handle “tasks that are unsafe, repetitive or boring.” It will carry about 45 pounds of goods at a time, deadlift about 150 pounds, and carry about 10 pounds of weight with a single-arm extension.
So, not really a super-powerful robot that can physically overpower most human beings. It will employ the same technologies Tesla’s cars use for Machine Vision and AI for autonomous driving.
We kind of know what the Tesla Bot will look like, thanks to some random person getting on stage during the announcement in a leotard and dancing around in a very SNL-like performance that was more intended for humor than for any educational purpose. It was a very awkward way to announce an entirely new product category for a company known for batteries, cars, and solar panels.
There were some very basic renderings of the bot shown during the presentation that detailed things such as the number of actuators in various parts of the body and a screen that will take the place of where you’d expect a human’s face. Beyond that, we don’t know much about this robot at all.
We don’t even know what it will cost — and given that this is a Tesla product, it won’t be cheap, and it’s nothing that a regular person will have in their homes anytime soon. In Musk time, a year for a prototype is more like five years.
But let’s assume Elon and Tesla pulls this off. Let’s say he can start volume production in three years. Where ideally would we want to see these things show up? And what are the most realistic applications for them?
Robots are already used in specific environments, obviously in automobile, semiconductor, and consumer electronics manufacturing. Companies like Foxconn have been moving to robotic assembly lines for the iPhone for several years. Amazon uses robotics in its warehouses to stock and retrieve goods for shipping to customers.
I’ve seen some videos of these environments, and they are truly awe-inspiring from a logistics execution perspective. And yes, the freaky Boston Dynamics robots are targeted for specific applications such as industrial and military. So those applications already exist, but these are highly specialized systems that don’t look anything like human beings and are certainly not meant to interface with or be around people like you and me in homes, in public venues, or the average workplace.
None of these environments use a general-purpose humanoid type of robot — they are programmed to do repetitive things in control loops with a bit of machine learning thrown in. However, like the Boston Dynamics ones, some are more sophisticated and more adaptable to their environments.
The kind of robots we’ve seen around people most recently, at least in experimental or early adopter form, have been motorized serving trays at restaurants and bars or patrolling urban areas and parking lots and large retail environments to assist human security.
The Tesla robot is the first that I have seen that isn’t some cute demo platform like Asimo that Honda used to make, but is supposed to or can exist in environments around human beings.
So just off the top of my head, these are the sort of applications I see robots — although not necessarily Tesla’s — working in:
Surgical and telemedicine, assuming the hands/manipulators have enough finesse and skill to operate surgical tools, people aren’t frightened out of their minds from decades of watching Star Wars to have their surgery performed with one, even if a human being is behind the wheel somewhere. The technologies are already in use for things like eye, bariatric, vascular, and jaw surgery. This is really more of a telemedicine-type scenario where the robots are piloted by human beings, as a UAV.
Addressing the shortage of workers not just in the front of the house at restaurants but also in the kitchen itself, doing more simplistic prep and cleanup, and possibly prepare fast food items. Based on my discussion with industry colleagues who are involved in the space, I am aware that companies like Starbucks and Mcdonald’s are heavily investing in research on robotic technologies to prepare their menus. Nevertheless, I feel that back-of-the-house kitchen work is a stretch of the capabilities of a general-purpose humanoid robot. In this application, these are robots that are more like the kind you see in a factory, specialized in their design, in kitchens that are retrofitted for them, or designed from the ground up for them. In the concepts currently being prototyped, the kitchen itself is one big modular robot, with stations specialized in doing a specific task and built to purpose — such as a fryer, sushi roller, bread toaster, pizza crust extruder/roller, sandwich assembler, or burger cooker/flipper, with machine learning and vision being used to do product quality control. It’s more likely that we will see robotic automation in stores of new configurations where customers never contact a human, or the front of the house is minimal and reduced to technology enablers to make sure the machinery is working optimally or acting as facilitators. While we will certainly see these types of robot kitchens sometime in the future, I do not expect to see a Tesla robot in a kitchen designed for humans and performing food preparation.
Working in law enforcement in strictly a patrol capacity in specific environments. Now, this is where we can start to get into dystopian, scary sci-fi areas. I think you could see them in shopping malls, in outdoor sports and concert venues, airports, that sort of thing. In these environments, a robot performs a similar function to a drone (another autonomous or piloted technology that we are starting to see heavily in use not just in military and law enforcement but in other civilian applications). This would be an on-the-ground, first-person POV of the action; perhaps it will also interact with the people there audibly or vocally. Of course, I think there will be a desire of certain governments, maybe not a US federal or regional government, to arm these with either lethal or non-lethal weapons, and I think that is where I think we need to be able to draw the line.
Musk stated his vision for these would be to do dangerous, repetitive, or tedious tasks in a factory — more or less. So assembly line kind of jobs. He also stated that the company has no intention to enable them for military or armed use.
We need to understand the programmatic limits of what these things can do before worrying too much about the ethics of employing these things in different scenarios. I suspect that because we are talking about a Tesla product here, many of the systems will be very locked down. Modifying them to be able to say, carry or operate a weapon designed originally for human beings, or programming them to use them is going to be restricted somehow.
Since they will have the ML andcapabilities of the cars, I think their ability to process information will be closer to what the vehicles can do from an autonomous standpoint. And right now, we know that the autonomous driving feature on the Tesla cars leaves a lot to be desired.
If they have wireless networking capabilities, such as, they could have extended processing or intelligence run out of cloud datacenter, which would have a lot more power than the localized CPU. So who knows what kind of decision-making these robots are going to have, or if these will be more of a teleoperated device, like a drone with hands that can walk around.
Personally, I am hopeful that these robots will be put to good use, helpful, and their pricing will not be restricted to only the super-wealthy. Are you going to line up to buy one when they are released for consumer use? Talk Back and Let Me Know.