ROBNEWS
The startup's dexterous gripper for factory floors skips motors at the fingers so a sub grain of sand sensor can read pressure from a light brush to a firm press.
Tacta Systems' new factory hand is built around a deliberately weird choice: there are no electric motors at the fingertips. Instead, the company routes pressurized fluid through cables that actuate each finger, keeping the heat and mass that motors would add out of the hand itself. That mechanical choice is what lets the company squeeze a tactile sensor smaller than a grain of sand into every fingertip, a sensor that, by Tacta's own claim, can read pressure across a 250-to-700,000-pascal range, or roughly the difference between a light brush and a firm press (The Robot Report).
The result is TactaBot, a three-part stack the company frames as the "missing layer" of factory robotics. The five-finger, 15-joint hand does the physical work. A Skill Capture system records expert demonstrations so the hand can be taught by showing, not by hand-coded scripts. A Dexterous Intelligence model, the "AI" piece in marketing copy, sits on top and turns those demonstrations into policies the hand can run. CEO Vikram Pavate, who co-founded the company roughly three years ago after identifying dexterous manipulation as the gap in factory automation, calls dexterous hands the "holy grail," a line worth flagging as a CEO quote rather than an industry consensus (The Robot Report).
The mechanism is the part the wire coverage tends to skim. Each finger has three joints and can lift about 25 newtons, or roughly 2.5 kilograms (just over five and a half pounds), per finger. The fluidic tendons, the company's term for the cables-and-piston setup, mean the actuators live in the forearm rather than the hand. Pavate says this lets the hand last up to 30 million cycles, the kind of lifespan an industrial buyer would actually want to see before signing a purchase order. Tactile sensitivity, not raw strength, is the bet: the 0.1-degree Celsius temperature precision and the sub-grain sensor are pitched at tasks where the robot has to feel what it is doing, not just place it (Tacta Systems announcement; PR Newswire).
That framing puts Tacta in a specific corner of the robotics conversation. Humanoid-robot coverage from bigger labs has dominated the past year's headlines, and dexterous-gripper startups have been one of the quieter subplots in that coverage. Tacta's $75 million in disclosed funding (Tacta Systems funding announcement) and this week's launch are an attempt to claim that corner before the humanoid story swallows the rest of the category.
The honest counter-question is whether tactile fidelity is the real factory-floor bottleneck, or whether it is the easiest one to point at. Most deployed industrial robots already struggle with perception in cluttered cells, planning under part variance, integration cost with existing PLCs, MES stacks, and safety zoning, and the slow, expensive process of getting an operator to trust a robot next to a part that costs more than the robot itself. A pascal-precise fingertip is impressive engineering, but the falsifier is concrete: if Tacta's pilot deployments cannot show measurable yield or mean-time-between-failure gains against those softer blockers, the bottleneck was elsewhere all along and the fluidic-tendon mechanism has not changed the factory's math.
The capability numbers cited above, the 30 million cycles, the 0.1°C precision, the sub-grain-of-sand sensor, the 2.5 kilograms per finger, are company-provided through the CEO interview, with no independent benchmarking in the source bundle. Trade press coverage is dominated by press-release re-reporting (The AI Insider, Robotics and Automation News). The earliest sign that the mechanism is doing real work on a factory floor will be a named pilot, a yield number, and a maintenance interval that beats what a human operator or a vacuum gripper already delivers on the same line.