Galaxea Dynamics’ Kengo brings bipedal humanoid teleoperation into sharper focus

The reported 140 x 44 x 22.5 cm footprint and 40 kg weight suggest a compact humanoid form that is easier to place in indoor environments than larger, heavier machines. Its 5 km/h / 1.4 m/s speed, along with legged bipedal mobility, points to a platform built for human-scale movement rather than warehouse-style transport. The listed 3- to 5-year battery life and safety features such as force limiting, collision detection, and emergency stop further frame Kengo as a system intended for repeated human interaction, not just isolated lab testing.

Kengo’s arrival signals that humanoid competition is increasingly about control architecture and deployment realism, not just height, speed, or acrobatics. If teleoperation becomes the dominant bridge to useful humanoid labor, the winners may be the robots that make human intent easier to translate into stable motion inside buildings where people already work.

Humanoid robots are now being judged less on scripted demos and more on whether they can support real-time human control, balance correction, and usable manipulation in human spaces. Galaxea Dynamics has unveiled Kengo, its first fully bipedal humanoid robot, positioning it as a platform for embodied AI, service work, and research rather than a fixed-motion showcase.

Kengo matters because its design centers on full-body coordination rather than isolated arm work, with bipedal walking, dexterous hands, and a software stack built around ROS and a proprietary embodied-AI layer. That combination points to the industry’s shift toward teleoperation-first humanoids that can mirror human motion with enough fidelity to be useful in service tasks and remote inspection. Its reported ability to handle interaction-focused payloads, along with stair-climbing and fall-recovery claims, suggests the robot is aimed at practical motion continuity rather than stage-only agility.
Kengo is less about single-task automation and more about making human control feel physically present.

In category terms, the technical flow is straightforward: human motion input goes into AI model processing, which then drives joint actuation and balance correction across the body. Kengo’s published stack points to that pattern, with a Linux and ROS-based base layer, a proprietary embodied-AI layer, and a sensor set that includes an RGB camera, IMU, gyroscope, force sensors, and visual SLAM with LiDAR mapping support. For a humanoid, that matters because the robot has to turn human intent into stable movement while continuously correcting posture in real time.

A realistic deployment scenario for Kengo is remote inspection and assisted handling inside a factory or service facility, where a human operator can guide the robot through narrow spaces, stairs, and cluttered work areas. In that setting, the value is not full autonomy but preserving the natural timing of human hands and feet while the robot handles balance and force control. That makes Kengo more relevant to environments where direct human presence is costly or inconvenient, but tasks still require judgment and fine manipulation.

Kengo

Rotaku’s Domo pushes teleoperation-first humanoid research into a smaller, cheaper platform

Rotaku says Domo measures 90 x 30 x 25 cm and weighs 20 kg, a compact form factor that makes the platform easier to place in labs and indoor environments. The robot is reported to move at 1.5 km/h / 0.42 m/s, which is enough for measured human-guided trials rather than fast traversal, and it is described as having a 3 to 5 year battery life, force limiting, collision detection, emergency stop, and collaborative mode. Its bipedal legged design, plus visual SLAM and LiDAR mapping, points to a system intended for controlled indoor locomotion and research use.

Domo also points to a broader shift in humanoids away from fully scripted motion libraries and toward systems that learn from humans in real time. That matters because teleoperation-first platforms can generate training data, expose control failures earlier, and give labs a clearer path from demonstration to policy learning. If this category keeps moving in that direction, the winning robots will likely be the ones that make human guidance feel immediate, repeatable, and safe rather than merely impressive in a demo.

Humanoid robots are increasingly being designed around a harder problem than simple walking: how to keep balance while imitating human motion and handling objects in real time. Rotaku’s Domo is the latest example, with the company unveiling a compact bipedal humanoid platform aimed at research, teleoperation, and embodied AI development.

What makes Domo notable is not just its size, but the way the platform is framed for human-centered control workflows rather than pre-scripted motion. Rotaku positions it around full-body control, teleoperation, locomotion research, and manipulation work, with a software stack that includes ROS2, Python, C++, and APIs for developers. The category’s real differentiator is that Domo is built to capture motion fidelity, balance correction, and operator responsiveness in one system instead of treating them as separate problems. Domo is less about static demos and more about making human motion usable as robot input.

The system flow is straightforward: human motion input goes into AI model processing, and the robot then converts that into joint actuation with balance correction. In practical terms, that means a teleoperation setup or learned policy can guide the robot’s body while onboard sensing helps it stay upright and react to disturbances. Rotaku’s stated sensor suite, including RGB and stereo cameras, optional LiDAR, IMU, gyroscope, force sensors, and ultrasonic sensors, supports that loop by feeding perception and stability data back into control.

A realistic deployment scenario for Domo is remote inspection or assisted service inside human-centric spaces where a person can guide the robot from a distance. In that setting, the value is not full autonomy but the ability to send a humanoid body through tight, familiar environments while preserving human judgment for the hardest parts of the task. That fits the product’s stated focus on teleoperation and interaction-heavy work better than large-scale outdoor deployment.

Domo

Persona AI Unveils Persona Humanoid for Shipyard Welding and Industrial Tasks

The Persona Humanoid is built to a height of 175 cm with a weight of 85 kg, enabling it to carry tools, packages, and precision instruments while interacting safely with people. It moves at a speed of 5 km/h and features a battery life of 5 years, ensuring extended operational periods without frequent recharging. Equipped with RGB cameras, stereo cameras, LiDAR, and ultrasonic sensors, the robot perceives its environment with high accuracy for safe navigation and task execution. Its legged bipedal mobility allows it to walk across varied terrain, making it suitable for the uneven surfaces of industrial sites.

This development signals a broader industry trend toward teleoperation-first humanoids, where robots are designed to augment human workers in human-centric spaces rather than operate fully autonomously. Persona AI's focus on shipyard welding reflects a move away from pre-programmed motion libraries, emphasizing real-time imitation and balance correction as key differentiators. The shift indicates that the next generation of industrial robots will prioritize collaboration and adaptability in complex, high-risk environments. This approach addresses the skilled labor crisis by enabling robots to perform tasks that require both precision and human oversight.

The global challenge of skilled labor shortages in heavy industry has prompted a major shift toward robotic assistance, with Houston-based Persona AI announcing the development of its Persona Humanoid specifically for shipyard welding and fabrication. Persona AI revealed a memorandum of understanding with HD Hyundai and Korean partners to deploy these robots for complex welding tasks, targeting prototype delivery by the end of 2026 with commercial field testing beginning in 2027. This move marks a significant milestone as the company expands its operations at the Ion in Houston to accelerate the creation of purpose-built industrial humanoids.

The Persona Humanoid stands out by prioritizing full-body coordination and real-time imitation fidelity over pre-programmed motion libraries, enabling natural interaction with human workers in demanding environments. Its design combines industrial-grade precision with human-like dexterity, allowing it to handle heavy tools for welding, grinding, and inspection while adapting to complex geometries without human input. The system is engineered to address critical labor gaps and enhance workplace safety, supporting supply chain resiliency in economically vital global industries. Persona Humanoid is less about raw hardware and more about redefining how humans control robots at scale through teleoperation.

The technical flow of the Persona Humanoid begins with human motion input, which is processed by an AI model to determine joint actuation and balance correction in real time. This architecture allows the robot to execute precise movements for tasks like welding and grinding while maintaining stability on uneven shipyard floors. The system integrates visual SLAM and LiDAR mapping for navigation, ensuring the robot can navigate complex industrial spaces safely. By leveraging ROS2 and proprietary software, the robot adapts to various workflows while maintaining high operational flexibility.

A primary deployment scenario for the Persona Humanoid is the execution of complex welding tasks in shipyards, where it works alongside human operators to tackle structural assembly and fabrication. The robot is designed to support skilled workers by performing repetitive, high-risk welding operations, thereby reducing physical strain and improving overall efficiency. Its rugged components and heat resistance allow it to operate in harsh settings, adapting to the unique demands of shipbuilding environments. This application highlights the robot's ability to handle precision tasks in physically demanding conditions.

Persona Humanoid

Zurich Startup Orbit Robotics Unveils Four-Armed HELIOS Humanoid for Space Station Maintenance

The robot is designed with dimensions of 160 x 60 x 40 cm and a weight of 85 kg, making it compact enough for tight station corridors yet robust for heavy lifting. It operates at a speed of 2 km/h (0.56 m/s) and features a battery life of 4 to 5 years total usable life, ensuring long-term reliability for extended missions. Equipped with RGB cameras, stereo cameras, IMU, gyroscope, force sensors, and ultrasonic sensors, HELIOS can detect collisions and limit force to ensure safety during close interactions with crew members.

This development signals a decisive shift in the humanoid industry toward teleoperation-first systems that prioritize human-assisted service over fully autonomous pre-programmed actions. Orbit Robotics, an academic spinout from ETH Zurich, chose a different approach by focusing on four arms rather than legs, highlighting that the future of space robotics requires specialized tools for specific environments. The market is moving away from trying to replicate human walking in space and toward leveraging human-like dexterity for tasks that astronauts cannot perform efficiently.

Space stations demand constant maintenance that steals valuable time from scientific research, prompting the need for specialized robotic assistants. Orbit Robotics, a Swiss startup based in Zurich, has unveiled HELIOS, a four-armed humanoid robot designed specifically to operate inside space stations and handle repetitive cargo and maintenance tasks. The robot was revealed in coverage published on May 21, 2026, with reports describing it as a space-focused machine built to move, stabilize, and work without relying on legs for navigation in microgravity.

HELIOS addresses the core challenge of real-time full-body motion imitation and balance in microgravity by utilizing four arms to hold position and manipulate objects simultaneously. Its key differentiators include full-body coordination for advanced multi-tasking, natural interaction with human crew members, high-fidelity real-time imitation, and robust teleoperation capability via the IKARUS platform. This design shifts the industry focus from pre-programmed motion libraries to teleoperation-first humanoids that can adapt instantly to complex station interiors. HELIOS is less about hardware and more about redefining how humans control robots at scale in space environments.

The technical system flow follows a clear path where human motion input is captured and sent to an AI model for processing, which then directs joint actuation and balance correction. Visual SLAM navigation allows the robot to map its surroundings dynamically, while the IKARUS teleoperation platform enables remote operators to guide the robot with precision. This input-to-processing-to-output architecture ensures the robot can stabilize itself against station walls and perform delicate tasks like satellite servicing without drifting away in zero gravity.

A primary deployment scenario for HELIOS is space station maintenance, where it targets the 35% of crew time currently spent on repetitive upkeep tasks that cost approximately $140,000 per hour. The robot uses its four arms to navigate station interiors, hold position against walls, and handle tools or precision instruments without relying on legs for mobility. This capability allows astronauts to focus on science while HELIOS manages cargo handling and routine repairs in the microgravity environment.

Robot News & Industry Updates

Galaxea Dynamics’ Kengo brings bipedal humanoid teleoperation into sharper focus

Latest Articles

Rotaku’s Domo pushes teleoperation-first humanoid research into a smaller, cheaper platform

Persona AI Unveils Persona Humanoid for Shipyard Welding and Industrial Tasks

Zurich Startup Orbit Robotics Unveils Four-Armed HELIOS Humanoid for Space Station Maintenance

More Robot Stories