TARS Robotics demonstrated a humanoid robot capable of performing hand embroidery during a live event in Shanghai , according to information released by the company. The demonstration showed the robot threading a needle and stitching a logo using both hands, a task that requires precise control of movement and force as well as coordinated manipulation of flexible materials.
The company said the task involves sub-millimeter accuracy, adaptive force control, and sustained bimanual coordination, all of which are technical challenges that have limited the automation of similar processes in manufacturing. TARS Robotics presented the demonstration as an example of progress in automating long-sequence, fine-motor tasks that are common in areas such as textile work and complex wire harness assembly.
During the event, TARS Robotics Chief Executive Officer Chen Yilun described the company’s approach as a combination of data collection, artificial intelligence, and physical system design. According to Chen, the company’s SenseHub platform is used to gather real-world operational data, which is then used to train the TARS AWE 2.0 embodied AI model. The model is designed to learn control skills that can be transferred to the company’s robotic platforms, including its T-Series and A-Series robots, with limited divergence between simulated and physical performance.
Chen said this integration forms a closed technical loop that connects data generation, decision-making, and physical execution, and that the approach is intended to support scalable deployment of embodied intelligence in production environments and consumer settings.
TARS Chief Scientist Ding Wenchao stated that increasing volumes of data from SenseHub and continued development of the AWE 2.0 model have led to higher task completion rates across multiple test scenarios. He added that further expansion of data and model capabilities is expected to improve the robots’ ability to generalize across different tasks, with applications envisioned in industrial and household contexts.
TARS Robotics was founded on February 5, 2025. The company reported that its core algorithms have been deployed in practical applications within its first year of operation and that its robotic platforms have undergone ongoing performance updates. TARS stated that it has raised a $120 million angel funding round, followed by a $122 million angel-plus round, with investors including Lanchi Ventures.
