ToraOne

ToraOne, developed by PaXini (also known as PaXini Perception Technology), represents a significant advancement in multi-dimensional tactile humanoid robotics. As the industry's first multi-dimensional tactile humanoid robot, ToraOne integrates cutting-edge sensor technology, AI-driven perception, and dexterous manipulation capabilities into a wheeled, foldable platform designed for seamless integration into human environments. Standing at a dynamically adjustable height from 1.46m to 1.86m, it features a total of 53 degrees of freedom: 21 in the body (excluding hands) and 32 in the dual dexterous hands (16 per hand, with 13 active and 3 passive DoF each). This configuration enables human-like flexibility, with four-fingered bionic hands powered by hollow cup motors for precise, high-force control down to 0.01N at the fingertips. The core of ToraOne's architecture is its revolutionary tactile sensing system, comprising 2280 self-developed high-precision ITPU (Intelligent Tactile Perception Unit) multi-dimensional tactile sensing units across the body and hands, delivering 6840 to 7824 tactile signal channels. These sensors capture 15 types of tactile data, including pressure, shear force, friction, softness, temperature, and sub-millimeter deformations, enabling unparalleled sensitivity for delicate tasks like handling fragile objects or food preparation. Complementing this is a robust visual system: 5 high-definition RGB monocular cameras (3 on head, 1 per hand) and 2 RGBD depth cameras (1 on head, 1 on chassis), paired with LiDAR and laser SLAM for 360-degree environmental depth mapping, obstacle avoidance (up to 2cm height, 7-degree climb), and autonomous navigation at speeds up to 0.6 m/s. At the AI core is the proprietary VTLA-Model (Visual-Tactile Large Model), a visual-tactile multi-modal AI framework that fuses sensor data for real-time 6D object pose estimation, grasping planning, and environmental perception. This edge-computing-enabled model supports imitation learning of human tasks, rapid generalization, and adaptive behaviors without heavy reliance on cloud processing. Motion control leverages advanced impedance and force-torque algorithms for safe human collaboration, with arm repeat positioning accuracy of ±0.5mm and payload up to 5-6kg per arm (96.5cm reach). The modular design allows customization for specific applications, powered by a 40Ah battery offering 8 hours of runtime and 6-hour recharge. Real-world deployment history underscores ToraOne's maturity. Debuting at the 2024 World Robot Conference, it demonstrated synchronized dances and tactile demos. At iREX 2025 in Tokyo, ToraOne made its public food-handling debut, continuously preparing ice cream with precision, highlighting operational dexterity in dynamic settings. Further showcases at Hong Kong Electronics Fair and plans for CES 2026 affirm its transition from prototype to commercial viability. Targeted for logistics (picking/sorting), manufacturing (assembly), healthcare (tool passing, rehab), and services, ToraOne reduces human risk in hazardous tasks while boosting efficiency. PaXini's omni-modal data factory accelerates its learning, positioning ToraOne as a benchmark for embodied AI, with ongoing iterations like TORA DOUBLE ONE building on its tactile foundation. (612 words)