Giving robots a sense of touch: contact location, normal/shear force, slip, texture, sometimes temperature/vibration. The under-built modality and arguably the missing sense for dexterity. Market ~$4.0bn (2025) -> ~$8.4bn (2030), ~16% CAGR (Mordor); robotics ~38% of demand.
Approaches & leading products
- Optical / camera-based (GelSight-type) — a camera images a deformable gel; very high spatial resolution but bulky, latency, fragile. GelSight (commercial leader) + Meta’s DIGIT 360 (18+ modalities, research).
- Magnetic / Hall-effect — magnet on soft substrate over Hall sensors, tracks 3-axis force. XELA uSkin is the leader (just integrated into the TESOLLO DG-5F hand). Robust, multi-axis, scalable.
- Capacitive — mature (touchscreens), lower resolution, drift.
- Barometric / MEMS — off-the-shelf MEMS barometers as cheap force cells; MEMS held ~37.5% of the tactile market in 2024. Cheapest path to volume. Tacterion (DE, flexible), Contactile (AU, 3D force + slip).
- Biomimetic fluid (BioTac) — SynTouch is small (~16 staff) but alive (NOT acquired/discontinued, a corrected assumption); BioTac remains a reference research sensor.
The investable angle (the readout, not the transducer)
The bottleneck is shifting from the transducer to getting dense, high-frequency tactile data off the skin and processed at the edge fast enough to close a ~200Hz control loop. High-res skin generates a torrent of data; on-/near-sensor processing (neuromorphic, MEMS-integrated) and the analog-front-end/readout IC are where differentiation and value capture sit. This is the most semiconductor-adjacent angle in the whole robotics cluster, and the basis of the Tactile Sensing Silicon thesis. Ties to Mems Sensing Actuation (deal coverage) and Low Power Edge Compute.
Connections
Physical AI (robotics cluster hub) · Dexterous Manipulation & Robot Hands · Tactile Sensing Silicon · Mems Sensing Actuation · Robot Autonomy Destination · Robot Actuators (the muscle layer)