

AGIBOT A2-W is a wheeled general-purpose robot designed for flexible manufacturing scenarios, capable of performing tasks such as grabbing, placing, transporting, and plugging. It features self-developed embodied intelligence algorithms, with open interfaces and tools to facilitate secondary development. The operational skills of AGIBOT A2-W improve with data accumulation, becoming smarter and more efficient over time. The A2-W is equipped with bionic dual-arm design, high-precision force control, multiple depth and vision sensors, and a four-wheel-drive system, ensuring stable, flexible, efficient, and safe operations in complex environments. Its fully integrated design allows for out-of-the-box deployment, and it supports hot-swappable battery replacement to reduce downtime. AGIBOT A2-W can be applied in scenarios such as loading and unloading, terminal plugging, and logistics transportation, making it an ideal choice for enhancing production line automation and achieving flexible manufacturing.

Machine Dimensions
770(L) * 620(W) * 1630(H) mm
Machine Weight
230kg
Battery Capacity
2kWh
Charging Time
2h
Operating Temperature
0–45℃
Single Arm Load
5kg
Degrees of Freedom per Arm
7
Operating Height
0–2m
Runtime
5h, supports hot-swappable battery replacement

Bionic seven-degree-of-freedom dual arms support parallel and asynchronous operations, enabling mid-air relay and hand-off to easily handle complex workpieces and specific postures.
Four-wheel drive with zero turning radius and crab-walking capabilities. The waist can elevate and tilt, offering 22 degrees of freedom to fully cover the human working space.
Integrated design of chassis, arms, and perception system, supporting deployment and commissioning within a single day.
Leveraging embodied atomic capabilities like UniGrasp, Uni6Dpose, and UniPlug, along with 275T computing power, the system achieves millisecond-level real-time object recognition, pose estimation, and operational decision-making, dynamically adapting to flexible working environments.
With 3D model synthesis training combined with reinforcement learning, the time to adapt to new objects can be reduced to within hours.


Modular design of atomic capabilities allows for OTA iterative upgrades; the task skill library can be continuously expanded, becoming richer with use.
Embodied algorithms can be continuously optimized through data collection and training, becoming smarter and more efficient over time.
Integrated design for data collection, simulation training, and inference deployment supports efficient data acquisition and deployment.
360° Lidar, 4 AI vision sensors, and 2 six-dimensional force sensors combine to provide multi-sensor fusion perception, enabling millisecond-level detection and intelligent obstacle avoidance.
The arm collision detection system, 360° real-time environmental sensing, and redundant perception and control design ensure safety for both personnel and the environment.


Supports hot-swappable batteries with energy replenishment in minutes and autonomous charging, reducing manual intervention.
Built-in real-time task self-diagnosis and self-recovery mechanisms minimize unplanned downtime and maintenance efforts.
Use our ROI Calculator to learn how fast automation pays for itself. No data on hand? Talk to our team—we’ll walk through your use case and ROI together.