Tacta Systems launches dexterous robot hand for skilled manufacturing work

Tacta Systems last week unveiled the TactaBot, a robotic system designed to handle high-skilled manufacturing work that has traditionally required human workers. The system consists of three integrated components: the Tacta Hand, a Skill Capture data-collection system, and a Dexterous Intelligence AI model. Vikram Pavate, co-founder and CEO, explained to The Robot Report that the robotics industry has succeeded in moving objects with vision-only and conventional robots, but "the holy grail is to be able to augment, complement, and expand the potential to use robots to do high value work." He noted that some of the world's biggest workforce shortages center on skilled labor in manufacturing environments.

The Tacta Hand is a five-fingered, human-sized gripper with 15 degrees of freedom (three per finger). Each fingertip can lift up to 25 newtons of force—approximately 2.5 kg (5.5 lb.)—and the entire hand is built around Tacta's proprietary Fluidic Tendon actuation technology. Rather than using motors or traditional cables, this system moves fluids through cables to push pistons and move the fingers. According to Pavate, this approach allows for very thin fingers, generates no heat near the hand, and achieves "incredible stability and precision" that motor-driven hands cannot match. The hand is also more reliable than cable-driven alternatives; if one cable fails in a traditional gripper, the entire hand breaks down, whereas Tacta's hand can operate for up to 30 million cycles. The fingertip sensors are designed as replaceable consumables, acknowledging that while the mechanical components have a long cycle life, the sensors will eventually wear.

The core of Tacta's capability lies in its proprietary tactile sensors, which are smaller than a grain of sand and can recognize the full dynamic range of human touch, from 250 pascals to 700,000 pascals. The sensors also measure temperature with a precision of 0.1 degrees centigrade. These are not MEMS-based sensors; each has its own analog-to-digital conversion and operates at a 1 mm pitch. The Tacta Glove, used for data capture, features 256 of these sensors on each fingertip. Pavate acknowledged that today's AI models for robotics are severely limited by a lack of relevant real-world data: "There's no Shakespeare and the web and Reddit and The New York Times to train these models in the physical world, especially for high value work." The Tacta Glove captures force, motion, video, and temperature data as workers perform complex tasks on the factory floor, with all data streamed wirelessly to a local host device. Workers can deploy the glove immediately without training and wear it naturally while completing their work; every hour of work further compounds the robot's training dataset.

Tacta's approach uses an off-the-shelf, pre-trained AI model that is then refined using the tactile and motion-capture data collected by the glove for specific tasks. When deployed at a customer site, Tacta fine-tunes the model using its gloves on the customer's actual operations, significantly reducing the training time needed to learn new skills. Pavate noted that in the long term, he envisions this technology in nursing homes, hospitals, food preparation, and even scenarios like a world-renowned sushi chef or pianist wearing the gloves to capture expert technique. For now, however, Tacta is focusing on manufacturing, an environment where the company's founders spent many decades, particularly in electronics. With reshoring a priority for U.S. industry and robotics suppliers, Tacta sees strong opportunity in electronics manufacturing. The company expects to ship its first TactaBots in early 2027.