Hanson Robotics’ Grace remains a landmark humanoid nursing robot as healthcare adoption stays uneven

Grace is reported to measure 165 x 50 x 40 cm and weigh about 30 kg, a size that keeps it human-scaled enough for close-contact settings. The robot is also described with a chest-mounted thermal camera, RGB eye cameras, IMU, gyroscope, and force sensors, features that support awareness, balance, and safer interaction in tight indoor spaces. Its software stack is said to include a proprietary operating system with ROS compatibility and Python APIs, which points to a system intended for integration and controlled customization rather than general-purpose autonomy.

Grace also reflects a wider industry change: humanoid robotics is moving from scripted demo motion toward teleoperation-first and task-specific systems that can work around current autonomy limits. That is important because healthcare places a premium on safety, predictability, and communication, all of which are harder to deliver than a staged walking demonstration. For now, Grace is best read as a sign that the humanoid market is looking for practical service roles before fully autonomous general-purpose robots arrive.

Humanoid robots are still wrestling with the hardest robotics problems: stable full-body motion, safe interaction, and useful work in human spaces. Hanson Robotics’ Grace, developed with SingularityNET through the Awakening Health joint venture, was demonstrated as a healthcare-focused humanoid designed to support older adults, patients, and frontline staff with conversation, monitoring, and social stimulation.

Grace matters because it was positioned not as a factory machine, but as a socially oriented humanoid built for trust, face-to-face interaction, and bedside-style engagement. Its design combined a humanlike face, multilingual conversation, thermal sensing, and healthcare-oriented dialogue, which made the robot more suitable for assisted service than a screen-based assistant. That mix also highlighted a broader robotics shift: humanoids are being judged less by novelty and more by whether they can create natural, low-friction interactions in care settings. Grace is a test of whether humanoid form can lower the social barrier to robotic assistance.

Input comes from the person in front of Grace, plus sensor data from its cameras, thermal camera, and motion sensors. Processing is handled by the robot’s AI software, which is designed to interpret speech, context, and basic health cues before coordinating movement and responses. Output is the humanoid’s physical reply: conversation, guided interaction, facial expression, and limited movement for service tasks or bedside engagement.

The clearest deployment scenario for Grace is social and monitoring support in eldercare or hospital rooms, where staff time is limited and patients may be isolated. In that setting, the robot’s value is not replacing clinicians, but helping with conversation, temperature screening, and routine engagement that can be repeated consistently. That makes Grace most relevant in environments where a human-like presence may encourage cooperation and reduce the burden of repetitive face-to-face check-ins.

Grace

Waseda’s AIREC pushes humanoid care robots toward teleoperation-first nursing tasks

Official project materials describe AIREC as a humanoid built for domestic, nursing, and medical support, with reported use cases including repositioning, bathing, feeding, ultrasound support, biometrics, and cooking assistance. The robot is described as standing 154 cm tall, weighing 150 kg, and using RGB, fisheye, depth, tactile, force, IMU, and gyroscope sensors, which together support perception, contact control, and balance. Reported software support includes ROS2, a proprietary DNN, and Python APIs, while visual SLAM is presented as the navigation approach for moving through human-centric spaces.

AIREC also signals that the humanoid market is moving toward teleoperation-first systems, where human expertise is captured, guided, and extended rather than fully automated on day one. That is a more practical path for care and inspection work, because it lets engineers validate balance, force control, and physical interaction in stages while keeping a human operator in the loop. For the sector, the bigger message is that useful humanoids may arrive first as trusted assistants in tightly defined human spaces, not as fully independent general workers.

Japan’s aging population is driving demand for robots that can support hands-on care without replacing human judgment, and Waseda University’s AIREC project is one of the clearest attempts to answer that need. The team has publicly shown AIREC and its Dry-AIREC systems performing tasks such as repositioning a person, assisting with dressing, getting someone out of bed, acquiring biometric information, and demonstrating ultrasonic examination support.

AIREC stands out because it is not framed as a single-purpose machine but as a humanoid platform built for full-body coordination in human spaces, where balance, force control, and natural interaction matter as much as motion. The project also reflects a broader shift in robotics: instead of relying mainly on pre-programmed motion libraries, the system emphasizes learning from skilled human movement and adapting force in real time. That matters in care settings, where a robot must be precise enough to help, yet gentle enough to avoid causing pain or distress. AIREC’s real value is in turning human caregiving motions into teachable robot behavior.

The basic flow is human motion input, then AI processing, then joint actuation with balance correction. In the demos and project materials, cameras and tactile sensing help the robot observe posture, limb position, and contact conditions, while the proprietary DNN and related control software estimate movement and force so the robot can adjust its actions during the task. The output is not just a scripted gesture, but coordinated whole-body assistance that can be adapted for tasks like repositioning, dressing, or supporting a standing motion.

The most concrete deployment scenario is bedside care, especially repositioning a patient to reduce strain on caregivers and help prevent pressure injuries. Waseda’s demonstrations suggest the robot is being tuned to learn how much force to apply while handling shoulders, knees, and hand placement, which is exactly where humanoids face their hardest safety problem: physical contact with a person who cannot be treated like a static object. In that setting, the value of AIREC is not autonomy for its own sake, but repeatable, monitored assistance that can work alongside staff.

AIREC

Hanson Robotics’ Professor Einstein shows how desktop humanoids are being packaged for education and interaction

Reported specifications point to a small humanoid form factor around 35 x 20 x 15 cm, which is closer to a shelf-top learning device than a full-size robot. A weight of about 1.5 kg and a wheeled shuffling style would help keep it portable, while microphone, RGB camera, and IMU sensing support basic interaction and body awareness. The stated 3 to 5 year battery-life range and app-driven software stack indicate a product designed for repeated household use rather than heavy-duty mobility.

Professor Einstein also reflects an industry pattern that remains easy to miss in larger humanoid headlines: many early products are being designed around teleoperation-adjacent or interaction-first behavior, not fully independent decision-making. For the category, that means the commercial test is shifting toward whether a robot can reliably hold attention, guide a lesson, or support a remote human-led experience in a controlled setting. In that sense, Hanson Robotics’ desktop humanoid points to a market where usefulness is measured by repeatable human engagement, not just mechanical ambition.

Professor Einstein, built by Hanson Robotics, was presented as a compact humanoid designed to talk, gesture, and support science learning through interactive games and app-linked conversations. The robot sits within a broader push to make humanoids useful in everyday human-centered settings, and its debut framed a familiar challenge for the field: how to turn expressive robots into practical learning companions rather than display pieces.

What made Professor Einstein notable was not raw autonomy, but the way Hanson Robotics packaged interaction around a recognizable human figure, voice recognition, and educational content. Reported features such as facial expressiveness, cloud-connected answers, and tablet-based lessons were aimed at making the robot feel approachable for children while keeping the experience structured and task-focused. That matters because the humanoid market is still searching for products that can sustain engagement without demanding advanced robotics expertise from the user.
Professor Einstein’s real innovation is turning humanoid presence into an education product, not a lab demo.

The system flow is straightforward for a humanoid of this kind: human speech or motion input goes into onboard sensors and the companion app, the software interprets the request through cloud AI and proprietary mobile integration, and the robot responds through facial movement, voice, and body gestures. In practical terms, that means the experience depends less on navigation and more on responsive interaction, which fits its role as a desktop companion rather than a roaming machine. For a user, the result is a robot that can answer questions, play games, and react in ways that reinforce conversation.

The most concrete deployment scenario is science education at home or in a classroom, where the robot can anchor short lessons, quizzes, and brain teasers around a child’s attention span. Hanson Robotics positioned it as a personal genius for kids, which suggests a workflow built around guided prompts rather than open-ended autonomy. That kind of use case is important because it asks humanoids to serve as engagement tools first, and general-purpose helpers second.

Professor Einstein

Hanson Robotics’ Zeno returns as a social humanoid built for interaction, therapy studies and classroom use

Reported specifications place Zeno at roughly 56 x 30 x 25 cm, giving it a compact footprint for tabletop or floor-level interaction, while its small weight and slow walking pace suggest it is built for safe proximity rather than fast movement. The sensor package is described as including binocular 720p stereo cameras, eight microphones, a speedometer, a compass and an IMU, which together support face-aware interaction and navigation in structured environments. Software support for RoboWorkshop, Java and C++ suggests the platform is intended for custom behavior development rather than fixed demo scripts.

Zeno points to a broader industry shift in humanoids: the market is no longer only chasing robots that can walk, but robots that can be understood by people in real rooms. That favors teleoperation-friendly and interaction-first designs, especially in therapy, education, and supervised service contexts where consistency and social cues matter. For Hanson Robotics, the strategic message is clear: the next phase of humanoids may be judged as much by how naturally they fit into human routines as by how well they move.

Zeno, the humanoid robot from Hanson Robotics, is being presented as a compact social machine designed to interact naturally with people, with demonstrations and product materials highlighting conversational behavior, facial expression, and human-robot interaction. The robot has long been associated with autism research and educational use, and recent descriptions emphasize its role as a platform for social interaction studies rather than a general-purpose household assistant.

What sets Zeno apart is not raw strength or speed, but the way it combines expressive facial motion, stereo vision, multi-microphone audio capture, and programmable behavior in a body small enough for close human environments. That makes it relevant to researchers who care about how people respond to humanoid presence, not just whether a robot can move from point A to point B. Its value is in the interaction layer: eye contact, speech, and motion cues that are closer to social signaling than industrial automation. Zeno is a testbed for making humanoids readable to people.

The basic system flow is human motion or speech input, then onboard processing through Hanson’s software stack, then output through coordinated facial movement, legged motion, and voice-based responses. In practice, that means cameras and microphones help the robot interpret a person nearby, the software organizes the interaction, and the actuators and facial mechanisms deliver the visible response. The result is a humanoid that is meant to respond socially as much as mechanically.

The most realistic deployment scenario for Zeno is a supervised therapy or classroom setting, where the robot can be used to support social interaction studies and structured learning exercises. In that environment, the robot’s small size, expressive face, and programmable behaviors matter more than autonomy, because an instructor or therapist can guide the session while the robot provides a consistent social cue. That makes Zeno better suited to controlled human-centric spaces than to open-world deployment.

Robot News & Industry Updates

Build Your Vision with Expert AI Execution

Hanson Robotics’ Grace remains a landmark humanoid nursing robot as healthcare adoption stays uneven

Latest Articles

Waseda’s AIREC pushes humanoid care robots toward teleoperation-first nursing tasks

Hanson Robotics’ Professor Einstein shows how desktop humanoids are being packaged for education and interaction

Hanson Robotics’ Zeno returns as a social humanoid built for interaction, therapy studies and classroom use

More Robot Stories