Memo Robot Learns Tasks From Motion-Capture Gloves, Skips Teleoperation

Gloves Spark Autonomy Leap

Sunday Robotics has unveiled Memo, a wheeled home companion that masters household chores through data captured by its Skill Capture Glove, eliminating the need for direct human teleoperation. This breakthrough arrives as the startup reports shipping over 2,000 gloves to homes worldwide, amassing nearly 10 million real-world chore instances from 500 households. By translating human movements into robot actions via the Skill Transform pipeline, Memo achieves a claimed 90% success rate on complex tasks, positioning it as a pioneer in scalable, autonomous domestic assistance without expensive robot fleets or puppeteering.

Glove Data Transforms Skills

Memo's innovation hinges on crowdsourcing dexterity from everyday users, who don the glove to demonstrate routines like sorting laundry or clearing tables. This approach generates robot-ready data matching Memo's dual-gripper end-effector precisely, fostering intuitive handling of varied objects from dishes to socks. Unlike rigid scripted bots, Memo adapts to unstructured home chaos, performing multi-step sequences with human-like finesse. Sunday's method scales effortlessly, turning remote human efforts into a foundation model that equips Memo for genuine independence in daily living spaces.

ACT-1 Powers Horizon Tasks

At Memo's core lies ACT-1, a foundation model trained on zero robot data, enabling ultra-long-horizon activities that baffle traditional systems. This software stack orchestrates everything from navigation to grip adjustments, drawing directly from glove-captured motions for seamless execution. By skipping teleoperation, Memo avoids the pitfalls of expert-driven demos, instead building generalization from diverse home environments. Early tests show it navigating 100 feet, manipulating over 20 objects, and chaining nearly 70 actions autonomously, redefining reliability for prolonged chores.

Chores Gain Playful Ally

Memo tackles practical home demands with flair, from loading dishwashers with plates, cups, and scraps to folding laundry piles and brewing espresso shots. It tidies shoes, sets tables, and organizes spaces, all while navigating via 3D maps learned during training. Its cartoonish face, soft silicone shell, and customizable hats make interactions welcoming, especially for families. Beta testing kicks off in 2026 for select Founding Families, promising out-of-the-box skills that evolve with user homes, transforming mundane routines into effortless harmony.

Dexterity Fuels Home Harmony

Memo's wheeled base ensures passive stability for safe, fluid mobility at human-paced speeds, enabling extended interaction sessions without fatigue. A suite of RGB, stereo, and 3D depth cameras paired with IMU, gyroscopes, force, ultrasonic, proximity, and gripper touch sensors delivers precise environmental awareness, powering dexterous picks of household items like dishes or tools. Indoor visual SLAM and 3D mapping support adaptive navigation, while compliant control and collision detection allow gentle human interruptions. Proprietary OS with ACT-1 stack, blending Python/C++ and likely ROS2 elements, sustains 3-5 year battery life for reliable daily assistance.

Rivals Faceoff Table