A2 lnteractive Service Robot AgiBot

A2 lnteractive Service Robot AgiBot

Humanoid Configuration and Ergonomic Design
Designed with reference to human ergonomics, the robot’s height and proportions closely resemble those of a human, with a highly futuristic appearance, featuring elegant yet powerful contour lines.
Adheres to ergonomic principles, and the HRI (Human-Robot Interaction) design enhances the experience of natural human-robot interactions.
Features a universal humanoid skeletal structure and modular design, supporting customizable exterior options to meet diverse user scenario requirements.

Multi-Modal Interaction and Intelligent Experience
Powered by large language models, realistic synthesis, full-duplex conversation, and edge deployment, the robot delivers lively and engaging real-time dialogues.
Combines large language models and RAG to build custom enterprise knowledge bases for efficient marketing, business consultation, and guided presentations.
Multi-modal models enable smooth interaction, with accurate sound source localization, facial recognition, and lip-reading in noisy environments, achieving 96% accuracy and 99% face wake-up rate.
ActionGPT generates human-like natural movements based on voice commands and intent, supporting light operational tasks.

Autonomous Mobility and Reliable Walking
Utilizing HIMUS (High-performance Multimodal Mapping System) 3D-SLAM algorithm and VectorFlux planning & control algorithm, the system supports L4-level autonomous mobility with low latency andminimal drift, even in complex, dynamic environments.
The self-developed, mass-produced integrated joint features high reliability, lightweight design, and ultra-high slot fill rate winding technology, delivering peak torque of 512Nm and stable operation forthousands of hours.
Combined with RTMOF (Real-Time Robust Motion Framework) non-linear model predictive control, reinforcement learning algorithms, and an efficient dynamic planning engine, it quickly generates globally optimal tfrajectories while enhancing motion strategies, ensuring agile, reliable, and flexible movement in uncertain environments.

Multiple Safeguards, Ensured Safety
360° Lidar and six high-definition cameras provide seamless panoramic perception without blind spots. Combined with open-set semantic information, it supports the synchronized construction of dynamic and static semantic environments and targets, enabling intelligent obstacle avoidance.
Three-layer safety monitoring at the business,system, and hardware levels, with redundant backup between the computing and safety centers; and dual-path control from the safety center to the actuators, achieving PLd-level safety protection.

Convenient Operation and Minimal Maintenance
Supports multi-platform control via remote, smartphone, and computer.
The Multi-functional standby station enables transportatio standby, and charging, reducing the need for human intervention.
Supports both quick battery replacement and charging, adapting to various scenarios.