Eggie

Eggie is a cutting-edge wheeled humanoid robot developed by Tangible Robots, a California-based startup founded to make robots commonplace in everyday environments, particularly homes. Announced in November 2025, Eggie represents a prototype platform emphasizing three core pillars: dexterity, compliance, and whole-body control. This architecture enables Eggie to thrive in chaotic, contact-rich real-world scenarios like cluttered kitchens or living rooms, where traditional robots falter due to rigid designs requiring pristine conditions. Architecturally, Eggie features a mobile wheeled base for smooth navigation at up to 3 km/h, paired with an expressive upper body including anthropomorphic five-fingered bionic hands capable of precise manipulation (1-2 kg per arm). Standing at 160 cm tall and weighing 45 kg, it boasts 20 degrees of freedom overall, with advanced electric servo motors and plastic/composite gear trains housed in an aluminum frame with composite panels. Sensory suite includes dual 1080p depth + RGB cameras for perception, supporting low-latency (<200 ms end-to-end) vision-language-action (VLA) models. Powered by a Linux-based ROS environment likely running on an NVIDIA Jetson or similar embedded AI board, Eggie integrates Bluetooth and Wi-Fi connectivity, with IP20 ingress protection and 7-hour runtime per charge. AI capabilities are central to Eggie's design, leveraging a full-stack vertical integration of hardware and software. Training pipelines incorporate imitation learning from teleoperated demos, reinforcement learning for robustness, dexterous manipulation policies, VLA models for multimodal understanding (3D vision, touch, scale), and co-training at scale. Currently, demonstrations like wiping coffee spills, picking mugs, lint removal from clothes, and household chores are teleoperated (2x speed in some clips), but the focus is rapid iteration toward full autonomy. Real-world deployments have already captured data from hundreds of unique environments and thousands of 'in-the-wild' interactions, enabling sim-to-real transfer unachievable in controlled sims. This data moat supports general-purpose manipulation for tasks like cleaning, object delivery, and interactive service. Deployment history, though early-stage, underscores Tangible's pragmatic approach: prioritizing real-world learning over hype. Videos from November 2025 show Eggie in messy scenes—full shots of cluttered manipulation—highlighting compliance for safe human coexistence. Targeted markets include home assistance, exhibitions, research, and education, with a charming white design fostering approachability. As a US-nationality prototype (price ~$32,000), Eggie positions Tangible against giants like Tesla Optimus by focusing on scalable, manufacturable hardware-software synergy. Future roadmap emphasizes autonomy scaling, with ongoing improvements in manipulation precision and environmental adaptability. Eggie's blend of approachable form, robust AI, and real-data training heralds practical home robotics, potentially revolutionizing daily chores by 2026-2030.