Teaching Robots Gentle Touch: GenRobot Debuts Next-Generation Training Hardware

2026-11-13 companies

San Francisco, Friday, 13 November 2026.
GenRobot introduced hardware capable of measuring tactile forces as delicate as 0.1 N, solving critical physical data bottlenecks to accelerate safe humanoid robot deployment across industrial operations.

A New Era for Humanoid Robotics

On November 13, 2026, GenRobot, a robotics startup based in Valley, CA, officially introduced its DAS 2.0 series, a groundbreaking hardware platform designed to bridge the critical gap between human dexterity and robotic learning [1][2][4][5]. The system comprises two core components: the DAS Dex 2.0 tactile glove and the DAS Ego 2.0 head-mounted multimodal data-capture device [1][2][4][5]. This innovation arrives at a pivotal moment for industries grappling with structural labor shortages and an urgent need to deploy embodied AI solutions in real-world environments [GPT]. GenRobot’s platform is engineered to accelerate the training lifecycle for autonomous and dexterous robotic applications, enabling automation engineers and enterprise leaders to scale data pipelines for faster humanoid deployment in industrial settings [1][2][4][5].

The DAS Dex 2.0: A Leap in Tactile Sensing

Central to the DAS 2.0 system is the DAS Dex 2.0 glove, which redefines tactile sensing with a 21-degree-of-freedom (21-DoF) hand-motion representation powered by miniature magnetic encoders and kinematic calibration [1][2][4][5][6]. The glove integrates 16 tactile sensors with 980 sensing elements, achieving a minimum detectable force of 0.1 N—a critical threshold for capturing the subtle interactions required in tasks such as handling fragile objects or fine motor coordination [1][2][4][5][6]. GenRobot conducted a month-long test of Dex 2.0, verifying the consistency of tactile responses in its semi-flexible piezoresistive electronic skin and side-mounted biomimetic exoskeleton [1]. The glove’s design ensures head-hand temporal alignment at approximately 1 ms, corresponding to ~0.5 mm displacement at a hand speed of 0.5 m/s [1][6]. Demonstrations at CoRL 2026 highlighted Dex 2.0’s ability to capture motion, touch, and vision in real time, with dexterity tracking accuracy of less than 2 mm, enabling tasks ranging from fine finger manipulation to bimanual coordination [3][5][8].

DAS Ego 2.0: High-Fidelity Data Capture for Embodied AI

Complementing the Dex 2.0 glove is the DAS Ego 2.0 headset, a first-person multimodal data-capture device engineered to synchronize visual, motion, and tactile data [1][2][4][5]. The Ego 2.0 features two 5-megapixel primary cameras and four 2-megapixel auxiliary cameras, expanding the horizontal field of view from 270° to 340° [1][2][4]. This broader field of view, combined with a 4 mm nominal fingertip tracking accuracy, ensures high-fidelity capture of human movements [1][2][4]. The device supports synchronous 30 fps imaging and utilizes hot-swappable magnetic battery packs, providing 2.5 to 3 hours of operation per pack [1][2][7]. GenRobot claims the DAS 2.0 system can deliver an effective data volume of up to 1 million hours, a 10-fold increase from previous versions, positioning it as a scalable solution for advancing embodied AI [10][alert! ‘GenRobot’s scale claims are self-reported and not independently verified’].

From Data Collection to Real-World Deployment

The DAS 2.0 system is designed to address a critical bottleneck in robotic training: the scarcity of high-quality, real-world data that accurately reflects human dexterity and tactile interactions [1][2][4][5]. Traditional methods often rely on inferring tactile data from video, which fails to capture the nuanced forces involved in tasks such as grasping or manipulating objects [11][alert! ‘This is a widely acknowledged limitation in the field, but specific comparative data is not provided in the sources’]. GenRobot’s platform directly measures these forces, enabling robots to learn from human examples with unprecedented precision [1][2][4][5]. The company’s data scale claims—including 1 million+ hours of off-the-shelf data, 5,000+ real-world scenarios, and 20,000+ hours of collection capacity per day—highlight the system’s potential to transform robotic training pipelines [11]. Industry experts suggest that such hardware could significantly reduce the time and cost associated with deploying humanoid robots in sectors like manufacturing, logistics, and healthcare [GPT].

Industry Implications and Future Prospects

The introduction of DAS 2.0 comes at a time when industries are increasingly turning to embodied AI to address labor shortages and improve operational efficiency [GPT]. For example, tasks requiring delicate force control, such as picking up a paper cup without crushing it or handling irregularly shaped objects, have historically posed challenges for robots [9]. GenRobot’s Dex 2.0 glove is specifically designed to address these challenges by capturing both hand movements and physical contact with high resolution [9]. The system’s ability to measure forces as low as 0.1 N could enable robots to perform tasks that were previously unfeasible, such as assembling delicate electronics or handling fragile materials in pharmaceutical settings [1][2][4]. Looking ahead, GenRobot plans to scale human data collection to further advance embodied AI, though specific timelines or deadlines for these efforts have not been disclosed [1][2][11].

Sources

Humanoid robotics Embodied artificial intelligence