AI Engineer World's Fair 2025
Your Personal Open-Source Humanoid Robot for $8,999 — Jingxiang "JX" Mo, K-Scale Labs
Read the talk
A humanoid robot you can program, repair and train
K-Scale’s K-Bot and Z-Bot connect modular hardware, learned motion and a shared simulation interface to a practical question: how does a developer start building robot applications?
From a talk by Jingxiang "JX" Mo
Before you start: Basic Python familiarity is useful; reinforcement learning, robot control and simulation concepts are explained as they arise.
What would make a humanoid accessible to developers?
How can a developer experiment with a humanoid when the available machines are expensive and proprietary? Jingxiang “JX” Mo, a founding engineer at K-Scale Labs, introduces a platform intended to open hardware, software and machine-learning models together. Labor shortages, household work and off-world exploration motivate the project; they are application goals, not a list of tasks the robots already perform.
The ambition is general-purpose robotics that people can build on, rather than technology available only inside a few companies. At the time of the talk, K-Scale has roughly fifteen people in Palo Alto and is preparing to launch its robots. The product and delivery plans throughout the presentation describe that launch period.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
A learned controller that can take a kick
K-Bot is a 4-foot-11-inch humanoid with an aluminum body and a reinforcement-learning controller for locomotion. Mo describes it as built in five months, with sensors intended to support manipulation and other applications. Preorders are open during the presentation, with October delivery planned.
The physical demonstration is direct: someone kicks the robot, and its controller handles the disturbance. This makes the learned motion policy tangible before the training machinery appears later in the talk. K-Bot also supports VR teleoperation with hands, and Mo describes running different locomotion and manipulation policies trained in K-Scale’s own framework. The kick is a demonstration of disturbance handling, not a quantified durability test.
Affordability starts with simplifying the machine. Mo presents K-Bot as an approximately $9,000 development and research platform—the $8,999 offer named in the recording’s title. His contemporary comparison puts the next-cheapest research option, a Unitree robot, at roughly $40,000 and a UR5 arm at roughly $15,000. Those are his estimates: the relevant distinction is a development-capable configuration, not simply the lowest advertised humanoid price. Unitree’s specifications distinguish base and development configurations.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Open construction makes the robot replaceable in pieces
The proposed open release extends beyond application code: a bill of materials, CAD designs, electronics, PCBs, software and machine-learning models are all part of the replication promise. That scope matters because modifying a physical robot can require changing its mechanical interfaces and electronics as well as its controller.
Modularity gives developers several ways to adapt the same body:
- End effectors: replace a five-finger hand with a parallel gripper, a UMI gripper or another tool using the mechanical interfaces.
- Limbs: replace legs or arms to repair damage or adopt a new hardware revision.
- Head and compute: replace the head as newer NVIDIA compute becomes available.
The intended upgrade procedure is mechanical replacement of a module, rather than replacement of the entire robot. A Python/Rust SDK then provides the programming interface.
Mo describes package installation as the entry point for programming. For locomotion, developers could use NVIDIA Isaac Sim to train a PPO policy; K-Scale itself uses MJX. Vision-language models and language models could also run through cloud services when local execution is unsuitable, alongside vision-language-action policies.
The default arm has five degrees of freedom, with a seven-degree-of-freedom replacement proposed for research that needs it. Hardware changes are paired with planned model and software improvements through over-the-air updates, which Mo envisions rolling out roughly weekly.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
From printable Stompy to manufactured K-Bot
The disclosed component set includes MIT Cheetah actuators, IMUs, audio modules, displays and cameras. Mo states an upper compute figure of 250 TOPS, but does not disclose the full specifications or the accelerator configuration and numerical precision behind that figure.
He dates the K-Bot effort to October of the preceding year and reports progress into mass manufacturing, with another completed design still unrevealed. The lineage begins with Stompy, a full-size, 3D-printable humanoid, proceeds through a prototype and then reaches the version developed with manufacturing partners. The slide makes that progression concrete: Stompy—May 2024, K-Bot V1—October 2024, and K-Bot v2—February 2025. The product website is the destination for the upcoming launch.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
A smaller robot with the same application stack
What if a developer cannot spend $9,000 on a humanoid? Z-Bot is the smaller entry point: approximately 1.5 feet tall, originating in a hackathon project that gained attention on Twitter and WeChat. K-Scale is bringing it toward mass manufacturing as well, although Mo gives no Z-Bot price.
The useful connection between the two sizes is their shared locomotion and software stack. Mo’s example is a voice-chat application: code written for the small robot can also be used on the large one. Z-Bot has a locomotion policy and simulator support, so its role is more than being a miniature display model.
Google DeepMind’s robot-soccer work inspired the small robot’s direction: a humanoid that can run around and play. The printable version also created a community before the manufactured product. Mo estimates about 5,000 Discord members and a few hundred people who built the 3D-printed robot.
He dates the Z-Bot project to November, describes manufacturing and launch as upcoming, and invites people to try the machines at recurring hackathons. Like K-Bot, Z-Bot is intended to be fully open source.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Two policies, with different responsibilities
Autonomy is the next layer above the hardware. K-Scale’s proposed architecture separates high-level action selection from whole-body motion:
| Layer | Responsibility |
|---|---|
| Vision-language-action policy | High-level control |
| Reinforcement-learning policy | Whole-body locomotion |
Both parts are still under development. A Rust firmware and software architecture powers the robots beneath those policies.
The developer-facing goal is a reusable Python application: someone implements a particular use case and shares it so other robot owners can run it. Mo likens the resulting ecosystem to an app store. That requires tools spanning learning, execution and access to the physical machine, rather than an isolated model release.
The GPU-accelerated learning framework uses MJX for locomotion and manipulation training. This is where K-Scale trained the RL controller used in the earlier kick demonstration. Integration and fine-tuning of generalist policies such as π0 and NVIDIA GR00T are work in progress: the aim is to make useful new models straightforward to run on the robot.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
The same control interface in simulation and hardware
K-Scale OS, or KOS, provides the software framework and Python interface for programming in Python or Rust. Mo contrasts its intended package-and-connect experience with the setup burden he encountered in ROS 1 and ROS 2. His spoken installation command is pip install kos; the inspected KOS README instead uses the pykos Python client.
The connection target is an IP address. KOS Sim, the digital twin, exposes the same gRPC control interface as the physical robot. An application can therefore change its communication target without changing the interface it calls. Using the documented client shape, a small Python entry point can make that endpoint explicit:
python
import argparse
import pykos
def connect():
parser = argparse.ArgumentParser()
parser.add_argument("--ip", required=True)
args = parser.parse_args()
return pykos.KOS(args.ip)
if __name__ == "__main__":
robot = connect()
Pass the simulator’s IP while developing, then the robot’s IP when targeting hardware. That switches the communication endpoint; physical policy performance still needs validation.
Because the simulator is open source, developers can start programming without owning a robot. Together, the operating-system and learning layers provide the route from policies and VLA models to applications on the hardware.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Train, inspect, evaluate, then target the robot
The training walkthrough connects those layers into a sequence for learning locomotion and manipulation—walking, grabbing objects or even dancing. Its first step is deliberately small: clone the training repository and invoke its Python training module. Mo describes the walking example’s train.py as about 500 lines, runnable on a local GPU or a rented RunPod machine using MJX acceleration.
-
Start training. From the cloned repository, run the demonstrated entry point:
bash
python -m train -
Tune the task. Training runs through millions of iterations of the robot attempting the desired behavior. Reward functions are adjustable, so the developer can change what the policy is encouraged to do.
-
Inspect learning. Use TensorBoard to follow loss and reward during training.
Mo estimates one to two hours to train a walking policy, without specifying the GPU or exact training configuration.
- Evaluate in KOS Sim. Run the trained policy and inspect the actual behavior: is the robot walking, standing or picking up objects as intended?
- Target hardware after evaluation. Once the simulated behavior is satisfactory, change the target IP to the physical robot.
This puts a behavioral check between optimization and deployment. Reward plots show how learning progresses; the simulated rollout shows what the resulting policy makes the robot do.
Mo estimates that the workflow can get a robot working in roughly one-tenth the time needed to set up other training libraries, but supplies no controlled comparison. The concrete reduction in setup work comes from packaging training, observation, simulation evaluation and hardware access into a connected workflow.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Building the stack with a community
A team of roughly fifteen people covers hardware, software and machine learning with help from an open-source community. Mo reports around 5,000 relatively active Discord participants across several servers, public bounties and contributions encouraged by MIT-licensed software. That licensing description should not be extended to every artifact: individual hardware and software repositories have their own terms.
Recurring hackathons give contributors access to the machines and a place to try applications. The prepared talk closes with hiring needs in electrical engineering, firmware and machine learning, and a plan to launch the robots and software stack in two to three weeks. The audience questions then turn from the development platform to its practical operating limits.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Battery operation and wall power
The robot uses a battery that slots into place and clicks in. Asked about the trade-off between battery weight and operating time, Mo reports about two hours of walking in tests so far. Wall-power passthrough also allows the robot to operate while charging. He does not give a battery mass or quantify the weight-versus-runtime trade-off.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
A consumer ambition, with teleoperation as the current limit
Are the first customers factories or households? Mo contrasts Figure’s and Tesla’s factory-oriented strategies with K-Scale’s ambition to become a US consumer humanoid company, selling to people interested in developing robotics. An accidental Shopify launch brought unexpected purchases from people interested in household tasks, programming and research. Companies also expressed interest in building B2B applications, including food-related work like the earlier conference demonstration.
A follow-up about unloading a dishwasher exposes the distinction between an accessible platform and an autonomous appliance. Current use is largely limited to teleoperation, Mo says. His broader assessment that nobody yet has a fully working VLA is best read in the context of general household readiness, rather than the absence of task-specific VLA demonstrations. A navigation and VLA stack that could support clothes folding or dishwashing remains a hope tied to improving model capabilities, not a demonstrated K-Bot feature.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Why KOS instead of ROS?
The ROS question asks what K-Scale gains or gives up by building outside that ecosystem. Mo acknowledges ROS’s strengths in nodes and asynchronous communication. His architectural rationale is specific to this robot: a policy-based machine with relatively few sensors requiring complicated asynchronous coordination. That reduces the need for the communication machinery he associates with ROS.
There is also a developer-experience preference. After frustrating ROS 1, ROS 2 and Ubuntu setup experiences, Mo wants a robot that arrives able to stand or walk and can be programmed from the purchaser’s existing computer. The desired starting point is an operating machine, not an environment-installation project.
Asked which AI accelerators the robots use, Mo names Jetson Nano and AGX as planned K-Bot compute choices at launch. The answer specifies K-Bot, not Z-Bot, and does not identify exact accelerator variants.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Learned positioning, and the final Optimus comparison
VR is a preferred teleoperation method. K-Scale trains what Mo calls a pseudo-IK policy: an RL model moves the robot toward a desired position instead of directly calculating inverse kinematics. The operator moves their arms to guide motion and presses controller buttons to open or close the gripper. Mo says Z-Bot will also support teleoperation through the same software stack.
The final question compares K-Bot with Tesla Optimus. Mo describes Optimus as mechanically more powerful, citing its linear actuators, and associates it with factory work. His view is that this difference need not produce an equally large difference in the use cases available to someone buying a robot to program; he does not present a comparative task test.
For price, Mo repeats an Optimus figure of at least $60,000 as something he last heard, not a verified selling price. He closes by putting K-Bot at $9,000 before mass production. The practical proposition is access to a programmable machine whose hardware, control software and learning tools can develop together—even while dependable autonomous household work remains unfinished.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Resources
From the talk
Public robot repository containing mechanical and electrical designs and links to control and training components.
Robot firmware framework with Rust build instructions and a Python client interface.
Simulation backend exposing the KOS gRPC interface, with installation and example-client instructions.
Official documentation for MJX, MuJoCo's accelerator-oriented simulation implementation.
Physical Intelligence introduces its vision-language-action model and task-specific manipulation demonstrations.
Further reading
MuJoCo- and JAX-based reinforcement-learning framework for humanoid locomotion and manipulation.
Walking-policy tutorial covering training, TensorBoard inspection, checkpoint conversion and simulation evaluation.
Research on teaching miniature humanoids to play one-versus-one soccer through reinforcement learning and simulation-to-hardware transfer.
Read the complete timestamped transcript
- 0:00
[upbeat music] Hello, everyone.
- 0:15
My name is JX, and, uh, I'm a founding engineer at K-Scale Labs. We build open source humanoid robots from hardware to software to, um, machine learning models, and we build it especially for developers.
- 0:29
Yeah. So humanoids have been, have been getting a lot of hype recently. You saw the Tesla Optimus, you have Unitree robots, you have, uh, 1X and et cetera. Um, they're quite proprietary, and they're quite expensive.
- 0:40
And the humanoids are getting so much hype is because of very big problems like physical labor shortage, consumer household, and also like off-world, uh, off-world exploration and et cetera.
- 0:52
Yeah. So for us at K-Scale Labs, our goal is to really solve general purpose, uh, robotics for everyone and open sourcing the entire stack to the entire world, so everyone will be benefiting from this really, really useful technology instead of a few different c- uh, companies.
- 1:10
Yeah. And, uh, our team is about fifteen people in Palo Alto. Um, you could visit us any time. We're launching some robots in the next coming month and, uh, yeah, I'll be demoing some robots in the slides.
- 1:25
Cool. So we're currently working on two robots, uh, the K-Bot and Z-Bot. The K-Bot is a four 11, uh, humanoid robot that we made in the last five months.
- 1:35
Uh, yeah, so full aluminum body, runs a RL controller for locomotion, and it has a-- It's pretty sensor complete, so it can do all the really cool tasks that, um, the previous presenter was showing.
- 1:46
And this will also be one of the cheapest humanoid robots on the market and is ready for pre-order right now, deliver by October.
- 1:54
Yeah. So if you come visit, this is a demo we do. Uh, [laughs] you can kick the robot. Um, the controller is quite robust, and the robot is-- can take a lot of damage.
- 2:05
Um, on the robot itself, you also have-- Basically, you can do VR teleoperation with building hands. Um, I'll describe the modularity a bit more. But yeah.
- 2:17
So the robot's able to run a bunch of different locomotion manipulation policies running our own RL training framework.
- 2:24
So when we started building the K-Bot, we thought about, like how do we make human, humanoid scale in the future, and what makes it, what, what makes it possible for people to adopt humanoid robots?
- 2:36
So we basically designed this humanoid robot to be the simplest possible, uh, factor as possible, and it's gonna be the most affordable development research-grade humanoid robot at nine thousand dollars.
- 2:48
Uh, the next cheapest option is probably at forty K, which is the Unitree robot. Um, this is roughly the cost of most robot arms today. Like, if you buy UR5, that's I think fifteen K worth, et cetera.
- 3:02
So the entire robot is going to be open source. What that means is we're gonna have OpenBOM. Every single piece of the hardware, CAD design, electronics, PCB, software, uh, machine learning models will be fully open sourced, so you can replicate this, um, if you want to.
- 3:17
And it's also gonna be very, very modular, which means that, um, you can basically change out end effectors. We have different mechanical designs that you can easily s- make interface with our, uh, end effectors, and you can just take out, uh, the hand to, for a parallel gripper instead of a five, uh, five fingers hand, or you
- 3:36
can use whatever end effector you want to use, like the Wumi grippers or et cetera. And that means is we can also easily upgrade and also fix the robot.
- 3:45
So our goal with selling this to developers is that when we have new hardware updates, you can easily just rescrew the r- robot knee in with brand-new legs or brand-new arms and also, um, brand-new head.
- 3:56
So, you know, as compute improves, like you get new N- NVIDIA chips, you can easily just add a new head onto the robot. Um, yeah, we're also building-- We have built the entire Python/Rust SDK for people to use.
- 4:10
Um, if you come visit us, you can start programming this robot basically immediately. It's like pi-piping style package, and you can start working on it. And it's capable of running the latest state-of-the-art ML al- um, algorithms.
- 4:22
So in terms of locomotion, you can use like NVIDIA Isaac Sim, uh, to train a PPO policy. We use MJX, um, which I'll explain a bit later, but you can use all kinds of different frameworks.
- 4:34
You can also run like different VLMs, like language models on the robot, maybe not locally directly, but you can run it through cloud, or you can run, uh, VLAs and et cetera as well.
- 4:44
Um, some-- Uh, this robot, by default, it's gonna be five DOF arms, but you can easily interchange to a seven DOF arm. Um, so that will suit most of the research need, research labs need.
- 4:55
And we'll make continuous like model and software improvements with OTA, uh, rollouts. So every m- every week, basically, we'll make new changes to the software as we go.
- 5:07
Yeah. Uh, here are some more specs. I can't release the full spec yet 'cause we're launching soon. Um, but yeah, you see it uses, uh, MIT Cheetah actuators, uh, pretty standard components with like, um, IMUs, uh, just different audio modules, displays, cameras, um, and up to two hundred and fifty TOPS compute currently.
- 5:30
Yeah. And we, we really started this project in about October last year. So we've been moving pretty fast. We have brought this to, uh, ma-mass manufacturing, and we have a new basically design that we completed that I can't show you now, but you'll be able to see in two weeks.
- 5:46
Yeah. So we started off this, uh, the K-Bot is like the K-Scale Stumpy project, like a full-sized humanoid robot that's 3D printable. Then we moved to like a prototype, and we worked with different manufacturing partners to actually make the new one, uh, that you see.
- 6:03
Yeah, it's launching soon if you're interested, um, in the K-Bot. Uh, you can go to kscale.dev.
- 6:09
Yeah. I'll give you a second. [clapping] No, I'm not done. Wait, wait, wait. Whoa, that's only one robot. I'm [laughs] not even one-third finished. [laughs] Joking, joking. Yeah, and so what if, what if you can't spend $9,000 on a cool humanoid robot?
- 6:23
Um, what if your, like, wife or husband doesn't allow you to do it? Well, introducing the Z-Bot, which is, uh, one feet five, um, 1.5 feet, uh, humanoid robot that we also made at K-Scale Labs.
- 6:36
So this started from a hackathon project we did. Um, it became really popular on Twitter and also WeChat, and so we're bringing this robot to mas- mass man- manufacturing as well.
- 6:47
It runs the same locomotion and software stack. Like, that means you can basically program stuff in, uh, for the small robot, but you can also put it on the big robot.
- 6:57
So, you know, if you make, like, a voice chatting app, you can just put it on the either robots, and it runs also a locomotion policy as well. It works out with all the simulators.
- 7:06
Um, yeah, we really got inspired by the Google DeepMind's, uh, robot soccer, robot soccer paper, where, you know, it runs around play soccer. That's really how we were envisioning it.
- 7:17
So yeah. Um, we have a pretty good-- The launch went really well for the 3D printing one, so our Discord has about 5,000 people. I think a few hundred people have actually made a 3D-printed one on the, the orange one on the bottom.
- 7:34
Yeah, so we also started this project in November, and we are already bringing it to mass manufacturing, which we'll also be launching very soon. Um, some people-- Yeah, we also run, like, monthly hackathons, so you can just come try out the robot.
- 7:47
Yeah. Yeah, same, same website. Okay. Okay, um, that's the hardware stuff. We talked about the hardware components we just open-sourced. Oh yeah, a- also, Z-Bot will be fully open-sourced as well.
- 8:01
Um, and so we also open-sourced our entire ML and software stack. So really, like, our core end goal is basically to make this, um, make the K-Bot autonomous. Well, so, you know, it's a pretty standard, uh, dual policy.
- 8:17
You have the high-level controller, which is a VLA, then you have the RL lo- whole, whole body locomotion policy. Yeah.
- 8:25
S- So what we really want right now is to basically finish-- We're currently working on both, basically, the RL part and also the VLA part. And we also made our own, uh, firmware/software architecture to power these robots in Rust.
- 8:39
Yeah, our end goal is basically to make the robot so easy to use. Um, any developer could, could write apps for robots. So, you know, Python application that you can re-share with people.
- 8:49
You make the robot do some very specific use cases that can be reused by other people. It's almost like a app store. And to do that, basically, um, we offer a lot of really cool developer tools we've been working on in the last six months.
- 9:03
So we open-sourced a library for basically GPU-accelerated robot learning. Well, it's mostly, like, locomotion manipulation training. Uh, we use MJX for this. And yeah, the vi- the video of you saw us kicking the robot, it runs the controller, RL controller, uh, R- RL model that we trained in this, uh, training framework.
- 9:25
Yeah, we also are working on to basically being able to integrate and fine-tune all the different VLA and generalist policies that you see from Pi Zero, also NVIDIA Groot, that we're also presenting today.
- 9:36
So this robot will be able to run, you know, we're trying to make the infrastructure very easy to run any cool models that you see that will be useful.
- 9:47
Yeah, we also made this, uh, operating system, which is like a software framework plus like a Python interface that you can use to program the robot using Python or Rust.
- 9:56
So, you know, you can just, instead of-- I don't know if you guys use ROS 1, ROS 2, they're pretty hard to set up. But using our system, you can just install Python package, like pip install QOS, and you can start programming a robot.
- 10:06
You just connect to IP. It's very, very easy to use. And we also have a digital twin in simulation. We call it a QOS Sim. It has the same gRPC interface you can use for, uh, controlling a robot in simulation.
- 10:19
And all you have to do between programming something in simulation and real is by changing the IP address. So you can prototype really, really rapidly without having to worry about breaking the robot, which is very cool.
- 10:30
So yeah, this is also fully open-sourced. Um, you can try it today. You can actually program the robot, um, just by using QOS Sim.
- 10:41
Oh, I don't know what happened to the images. But yeah, and then what the, what basically what the machine learning and the operating system layers enables for us to run different policies, VOA models, and on our robot hardware, and for people to develop really cool applications with.
- 10:58
So yeah, I'm just gonna go through like a very, very quick, uh, RL training and deployment examples of how, um, researchers and developers could use our robot to train a local manipulation policy for the robot to, you know, grab different things or walk around or even dance.
- 11:15
So yeah, RL training setup is very easy. You just git clone the repository, and then all you have to do is run python dash m train. And in this train.py is you effectively have all the training code you need, uh, abstracted.
- 11:28
It's about 500 lines for walking. Yeah, and then basically you can run this on like, you know, using RunPod or your local GPU. It's MJX, so it's like, you know, it's accelerated, um, accelerated compute.
- 11:42
And training a walking policy roughly takes one hour to, uh, two hours. And yeah, so you're just gonna run through like millions of different, um, not examples, uh, yeah, iterations of the robot performing tasks you want, and you can tune the reward functions and et cetera.
- 11:57
And you can see the loss and reward functions in, um, our observability, basically TensorBoard. Um, yeah.
- 12:06
And afterwards, when the robot's finished training, you can easily evaluate it in CoSim. Uh, so all you have to do is like K- uh, K inverse N, like you run the policy, yeah, and in simulation, and you see the robot, if it's walking, if, well, it's, if...
- 12:22
You can see if it's doing the thing you want it to be doing. For example, like walking, standing, uh, picking up objects. And if that's really good in simulation, then you can just easily change the IP address, and then you have sim-to-real deployment.
- 12:38
Yeah. And, uh, s- it's very cool. Like you can basically get a robot to work in like one-tenth of the time of like what you would take to set up most, uh, [laughs] training libraries right now.
- 12:50
And also, like we're a team of 15 people, so we do everything from hardware to software to ML. So how we're able to do this is actually by working with our open source community.
- 13:01
Currently, we have about like 5,000 relatively act- act- active Discord members in a few servers. We have a lot of public open bounties that people are tackling. And of our, because of our software's MIT licensed, yeah, a lot of people are coming to help us.
- 13:16
And we also run hackathons almost on a bimonthly basis that a lot of people come to participate.
- 13:23
Yeah. So we are also hiring electrical, firmware, and ML engineers, so if you're interested, feel free to, uh, ask me and then also go on the kscale.dev/joint website. Um, yeah, we're trying to hire a lot more crack people to join us. [laughs]
- 13:42
Yeah, so we're launching the robots and software stack in about two, three weeks. Um, if you're interested, follow on the website. I'll be ha- happy to answer any questions. [audience applauding]
- 13:57
Yes.
- 13:57
A little bit is fine.
- 13:57
Yeah. Yeah, sounds good. Uh, go ahead. Yes.
- 14:02
Where's the power? Where's the power? The battery? Yeah. Is it, is it battery packed? Yeah, yeah, it's battery packed. Yeah. It just has a battery. I don't know if I can show you in the picture.
- 14:11
Yeah, it's behind this, basically. You can just slot it in, like it clicks in. Yeah.
- 14:17
So it was very... Did you talk at all about trademark, the trade-offs between, um-
- 14:22
Mm-hmm
- 14:22
... the, the, the weight of the battery and how, what the longevity of these things are?
- 14:27
Yeah, the weight of the battery versus longevity? What do you mean?
- 14:31
Yeah, longevity on a char- like a charge.
- 14:32
Oh, like how long it is?
- 14:34
Yeah, like how long does it last?
- 14:35
Yeah. Uh, walking so far our test is about two hours, but you can pass through, so you can po- uh, power through the wall plug. So it's, yeah, you, you can just keep letting it charge and also run it at the same time.
- 14:47
Yeah. Yes. Uh, well, bl- black jacket. Yeah. [laughs]
- 14:56
What are the use cases that you guys are imagining to start off with?
- 14:59
Yeah.
- 14:59
Is this gonna be more for like commercial, like, you know, factory kind of use cases, or do you envision this mo- more in the home, like helping out?
- 15:06
Yeah, so basically our bet, so a lot of companies like especially in the US are betting on like B2B. So like Figure, for example, are selling to factories. Seems Tesla itself is the customer.
- 15:17
For us, our really bet is be- becoming the first US consumer robotics company, like a robotic- a hu- humanoid robotics company. Yeah, so really selling it to anyone that's interested in developing robotics.
- 15:29
So a lot of our current customers, we accidentally launched our r- robots. Like people accidentally start- started buying our robots through our Shopify page. Um, that was a complete mistake, but a lot of people that bought were just people genuinely interested in using for household tasks, like programming, uh, for different research, and also there are a lot
- 15:48
of companies also interested in working with us, um, to make B2B businesses. Like for example, the food, um, food demo that we just saw. Yeah.
- 15:57
Sorry if I can ask a follow-up question.
- 15:58
Yeah. Keep going. Yeah.
- 15:59
Like what household like chores do you think it would be well-suited for? Like-
- 16:03
Yeah, um-
- 16:03
... unloading the dishwasher, for example?
- 16:05
Yeah, yeah. I mean, right now, right now we don't have any, I don't think anyone really has a fully working VLA model yet. So right now it's pretty limited to teleoperating system, uh, sorry, teleoperation.
- 16:16
So yeah. Um, but soon we sh- we hope to be able to have like this navigation VLA stack for you to do like, you know, folding cloth or like doing dishwashing as the model capabilities improve.
- 16:27
Yeah. Yes. Oh, uh, white, white shirt. [laughs] Oh, I mean green sh- yeah, you can go first.
- 16:35
Can, can I ask a-
- 16:36
Yeah, yeah, of course. Yeah.
- 16:36
Okay. Yeah. Uh, so the, um, you kind of alluded to complexity of ROS 2 in terms of the setup.
- 16:43
Uh-huh.
- 16:43
Uh, I'm wondering if there were other like benefits and trade-offs that you considered for foregoing something like ROS and ROS 2 in that ecosystem.
- 16:51
Oh, yeah. Like why not use ROS basically?
- 16:53
Yeah.
- 16:54
Yeah. Well, there are a lot of reasons. Um, our robot is mostly programmed. So ROS is really good 'cause like the nodes and stuff, right? So like the asyn- like the communication, but for our robot, we really don't have that many sensors, and we really wanna do this like model-based ro- like policy-based robot.
- 17:09
So we don't have many complicated sensor that we need to like async communicate at all times. Um, the other part is like we're, I'm, I'm [laughs] pretty opinionated. I used ROS 1 and ROS 2, Foxy and, uh, Noetic.
- 17:23
Um, I've just had a pretty bad experience using it, having to set up Ubuntu, you know. Like I just want a robot I can just buy, uh, buy, open the box, it stands or walks, and then I can just start programming it using my computer.
- 17:36
Yeah. Yeah, yeah.
- 17:40
What k- what kind of AI acceler- Yeah. What kind of AI accelerator is on each of the robots?
- 17:46
Yeah. So, um, basically for the K-Bot, it's gonna be Jetson Nano and AGX. Yeah. So yeah, there are different compute options you'll be able to choose when we launch.
- 18:02
Yes. Yeah, go ahead.
- 18:03
How do you teleoperate it?
- 18:05
Oh, so you can just put a either VR headset. So there are a few, few different methods. So the preferred option for a lot of people is VR headset.
- 18:12
We have this like, we also train like a pseudo IK. So basically it's like a b- go into position using like a RL model instead of just like calculating IK, but it works pretty well with our VR setup.
- 18:25
So you can move the hand gestures, so you can click button to open and close gripper and just, yeah, move your arm and stuff. Yeah.
- 18:32
Can you teleoperate the small one too?
- 18:33
Yes. Yeah, you'll be able to. It runs the exact same software stack. Yeah.
- 18:37
Last question.
- 18:38
How does it compare with, uh, the Tesla Optimus? The Tesla humanoids? Uh, in terms of like mechanical powerness, like, you know, the Tesla's way more powerful. It has like linear actuators and et cetera, but in terms of like actual use cases, I don't think it's really that different.
- 18:55
Yeah. I mean, Tesla is actually built for like a factory type of, uh, use cases, but in terms of like you wanna, for people to actually buy and use this robot, it's not very different.
- 19:06
What's the price difference? Uh, I think the last time I heard Tesla Optimus is about 60K at least. Um, we can ask some Tesla engineers. [laughs]
- 19:16
But, uh, our robot's $9,000 before mass production. [upbeat music]