California-based robotics startup Satyress is developing Threehalves, a 7-foot-tall teleoperated centaur robot designed for hazardous work.But if it’s designed for hazardous tasks, why does it look like something you could ride—something that seems more at home in an amusement park? Although its appearance is a bit creepy lol.
Four-legged robots can reach many places that wheeled vehicles cannot. Isn’t it a bit of a waste to treat them as mere means of transportation when they can carry loads weighing several metric tons? Their current speed is still a bit too slow—is this due to safety restrictions, or are the hardware limitations—such as the motors and batteries—preventing them from going faster?
I have to admit, this paper is technically brilliant and also made my skin crawl a little.
They built a tiny oxygen-generating diving suit for a cockroach, implanted electrodes to control its movement, and kept it walking underwater for hours.
The engineering is genuinely impressive. But the more capable these cyborg insects become, the harder it is not to feel uneasy about the living animal underneath all that hardware.
For aesthetic reasons, robots are typically equipped with outer casings around their core components. Could the degrees of freedom of a bare robot be compromised in terms of performance due to design constraints? Is that why the demonstration robot in the example uses rubber gloves as its outer casing? Is this a better choice?
I've noticed that in many demos of embodied robots, their movements are still quite slow.
Is this mainly because we don't have enough real-world data to train good control policies yet, or is it more of a hardware limitation, such as motors, actuators, power, or thermal constraints?
It also seems that visual tracking can still lose fast-moving objects pretty easily. Does perception latency or tracking reliability become a major bottleneck for how fast robots can move in the real world?
Curious how people in robotics think about this: is the main limit data, control algorithms, hardware, perception, safety, or something else?
I’ve noticed that a lot of humanoid robot demos, especially for home robots, use laundry folding as a showcase task.
I’m curious why this task is so common. Is folding clothes considered a particularly good benchmark because it involves deformable objects, perception, dexterity, and planning? Or is it mainly used because it is an easy-to-understand household chore for the general public?
Hi everyone, I’m interested in the HCI/HRI side of embodied robots in education, outside the usual coding-club or robotics-team context.
A lot of examples I’ve seen are about teaching children programming, robotics, or basic STEM skills. I’m wondering whether there are convincing use cases in regular learning environments, such as science classes, museums, special education, language learning, or informal learning spaces.
From an HCI perspective, I’m curious where physical embodiment actually changes the interaction, rather than just adding novelty.
For example, does a robot help with attention, social presence, embodied demonstration, motivation, collaboration, accessibility, or teacher facilitation?And on the other side, where do these systems usually fail: novelty effects, classroom management, maintenance, safety, cost, limited curriculum fit, or unclear learning outcomes?
I’d love to hear from people who have studied, designed, evaluated, or deployed educational robots. Are there papers, projects, or field studies you think are especially useful for understanding what actually works?
I understand the basic argument for humanoid robots: they can interact more naturally in environments built for humans, and companies can use human data to train them, which may lower costs.
But if the goal is simply to perform a specific task, wouldn't other shapes often be more efficient?
It feels like there is so much room to explore non-humanoid robots instead of trying to make everything look and move like a person.
