Tangible Robots Unveils Eggie, Bipedal Humanoid for Home Dexterity Tasks

Eggie Emerges from Stealth

Humanoid robots struggle with real-time full-body motion imitation, balance, and manipulation in unstructured home environments. San Francisco-based Tangible Robots emerged from stealth this week, unveiling Eggie, a legged bipedal humanoid designed for dexterous household interactions, as demonstrated in a video where it cleans a coffee spill by grabbing a mug and wiping the counter. The startup, founded by MIT PhD dropout Bipasha Sen and backed by $4 million in funding, emphasizes Eggie's ability to thrive in human-centric chaos through in-house hardware and AI.

Dexterity Redefines Interaction

Eggie stands out with full-body coordination for natural interactions, real-time imitation fidelity from real-world data training, teleoperation capability, and compliant arms that enable safe contact. These features address the core challenges of manipulation in messy spaces, unlike rigid industrial bots. Eggie prioritizes teleoperated compliance over pre-programmed routines.

Motion Pipeline Explained

Human motion input, captured via sensors during data collection with exoskeletons, feeds into AI models processing vision, touch, and balance data. This drives joint actuation with real-time corrections from the IMU, gyroscope, and force sensors for stable bipedal walking. Visual SLAM navigation ensures precise movement in dynamic homes (Demonstrated in controlled conditions).

Laundry Folding Unfolds

In a teleoperated household service scenario, Eggie folds laundry by imitating operator gestures: dual 1080p cameras detect fabric folds, compliant arms apply gentle force to avoid tears, and bipedal mobility navigates around furniture. This assisted task leverages its payload handling for fragile items and clothes, reducing physical strain on remote users in elder care setups. Real-world training on thousands of interactions enables adaptive gripping for varied textiles.

Specs Enable Home Mobility

Eggie's 160 x 50 x 40 cm frame and 45 kg weight support bipedal stability at 3 km/h speeds, ideal for indoor navigation (Manufacturer claim · Not independently verified). A 3-5 year battery life powers extended sessions, while Linux-based ROS and Python APIs facilitate custom teleoperation. Safety via force limiting, collision detection, and emergency stops ensures compliant people interaction.

Rivals Edge Check

Teleop Shifts Humanoids

This launch underscores a pivot in humanoids toward teleoperation-first designs, prioritizing operator control in human spaces over full autonomy. Unlike motion-library bots, Eggie's data-driven imitation scales with human input, hinting at hybrid deployments in assisted living. Investors like Blume Ventures back this as a pragmatic path to commercialization.