6thSense builds wearable tactile sensing rigs that capture how human hands interact with the world — pressure, contact, grip, and pose — across touch, vision, depth, force, and 3D hand pose, all in real environments. We sell that multimodal data to the companies training the next generation of dexterous robots.

https://www.youtube.com/watch?v=gXx53tcIv5U

Robots are getting good at seeing. They're still terrible at touching. That's because there's almost no data capturing what a hand actually feels while doing real tasks in real homes. Lab data doesn't cut it. We go anywhere and capture the missing layer that manipulation models need to learn fine motor control.

What makes our data different is the hardware we built to collect it. Our tactile sensing devices are designed and manufactured in-house, purpose-built to capture the richest touch signal in the industry — with scale in mind from the very beginning, enabling us to actually solve the tactile data problem. Every sensor maps directly to how robotic skin needs to work: a 1-to-1 analog for the tactile arrays going into the next generation of robot hands. That means the data we collect today doesn't just train better policies. It transfers directly to the hardware it's meant to run on. The team comes from Tesla, Mecka, Samsung, DoorDash, and Amazon, with expertise across hardware, sensors, ML, robotics, and software at scale. If you're training a manipulation policy or need custom tactile data, reach out.