Sony Patents a Robot That Uses Touch and Distance Sensing Together to Understand Objects
Most robots rely on cameras or lasers to figure out what's in front of them. Sony is filing for something that also feels its way around, combining distance data and touch readings before deciding what to do next.
How Sony's robot combines touch and distance to "see" objects
Every time a robot arm reaches for an object on a shelf, it makes a quiet bet: it trusts that what it sees is what it will grab. When that bet is wrong, the robot drops things, knocks them over, or freezes entirely.
Sony's patent describes a control system that hedges that bet. Instead of relying on a single sensor, the robot simultaneously reads how far away an object is using a distance sensor and what the object actually feels like using a touch sensor built into its body. A recognition system stitches those two streams of information together into a clearer picture of the object before the robot takes action.
The result is a robot that doesn't just see an object; it builds a richer understanding of it before it moves. That matters most in situations where vision alone misleads, like objects that are transparent, oddly shaped, or partially hidden.
a recognition section that recognizes a measurement target on a basis of distance information about the measurement target measured by a distance measurement sensor provided on a robot and tactile information about the measurement target measured by a tactile sensor provided on the robot …
Translation: The system identifies objects using both sight or distance data and physical touch sensors on the machine.
How the recognition section fuses sensor data into control signals
The patent centers on three connected pieces:
- Distance measurement sensor: mounted on the robot, this measures how far away a target object is, typically using laser or infrared signals (the kind a self-driving car would use to map its surroundings, scaled down).
- Tactile sensor: also mounted on the robot, this detects physical contact with an object, capturing information about texture, resistance, or surface shape.
- Recognition section: a processing layer that takes both data streams simultaneously and builds a unified description of the object.
The controller then uses that unified description to issue movement commands. The key claim is that recognition happens on the basis of both distance information and tactile information together, not one after the other.
This matters because the two sensor types cover each other's blind spots. A distance sensor can tell you something is 30 centimeters away but can't tell you whether it's rigid or soft, thick or thin. A tactile sensor tells you exactly what you're touching but only once you're already making contact. Fusing them gives the robot a richer model at the moment of interaction.
The patent doesn't prescribe a specific fusion algorithm, leaving that architecture open, which is typical for a foundational control-system claim.
What dual-sensor robots could mean for factory floors and homes
For robots doing physical work, misidentifying an object isn't a software glitch you patch remotely; it's a dropped glass or a damaged component on a production line. A system that cross-checks visual distance data against real touch feedback before acting reduces that failure rate in a direct, hardware-level way.
Sony's steady investment in robot sensing suggests the company sees physical robotics as a long-term product area, not just a research showcase. For consumers, the near-term relevance is in industrial or service robots rather than home devices, but the underlying idea, that robots need to feel as well as see, is one that will eventually show up anywhere a robot handles something fragile or unpredictable.
That makes this Sony's sixth filing we've tracked since July on teaching robots to grasp and move, building on earlier applications covering self-correcting robot grip and faster obstacle dodging.
Adding a touch sensor alongside a distance sensor means more parts that can break, more data to juggle in real time, and a higher price tag on every unit. That cost is easy to justify when a robot is handling objects that are predictable in category but vary in exact position or surface texture, like parts on an assembly line.
The harder question is whether touch sensors can keep up in messier, more varied environments. Calibrating a sensor to reliably read a rubber gasket, a glass jar, and a wet surface are three different problems, and the patent says nothing about how the two data streams are actually combined into a single reliable picture.
That silence is strategically useful for Sony but leaves the most fragile part of the design unresolved. The concept is sound; whether the final product holds up depends entirely on engineering work the patent does not describe.
There are more where this came from
We read every patent application Big Tech publishes and send you the ones worth knowing. Plain English, free, every week.
The drawings
41 drawing sheets from US 2026/0264250 A1 · click any drawing to enlarge
Want this weekly breakdown for a company we don't cover? Patentlyze Pro →