Robots That Grab Objects Without Knowing Their Exact Spot

Ocado's patent describes a robotic gripping system that can precisely manipulate objects of varying sizes and positions, even when the robot doesn't know the object's exact starting location.

What it covers

This patent describes a robotic device, system, and method for manipulating objects. The core idea is a "motorised linkage" (like a robot arm or gripper finger) that can be precisely positioned by a controller, even when the "precise location of the object with reference to the device is unknown" (Claim 1). This means the robot doesn't need to know exactly where an item is before trying to grab it. For example, if a robot needs to pick up a box from a shelf, it doesn't need a perfect 3D scan of the box's exact coordinates. Instead, the device can use "force and/or torque controllable" joints (Claim 2) to adapt its grip once it makes contact, ensuring the grip is stable regardless of the initial contact point. This allows the robot to handle "plural inventory items located in a warehouse" (Claim 6) that might be slightly out of place or have different shapes.

What it doesn't cover

• —Gripping devices that require a precise, pre-known location of the object before manipulation.
• —Robots that rely solely on vision systems to precisely locate objects before attempting to grip them.
• —Gripping systems where the relationship of the linkage to the object *is* dependent on the initial contact position.
• —Methods of manipulating objects without using a motorized linkage.
• —Systems that do not include a software utility for controlling device movement.

The clever bit

The cleverness lies in enabling precise object manipulation *without* requiring precise prior knowledge of the object's exact location. By making the grip adaptable through force/torque control once contact is made, the robot can compensate for initial positional uncertainties, simplifying the overall picking process.

Why it matters

This patent addresses a significant challenge in robotics: handling variability in object position and type. In environments like automated warehouses, items are rarely in perfectly predictable spots, and a wide range of products must be handled. This technology allows robots to be more flexible and robust, reducing the need for complex vision systems or highly structured environments. Ocado, as a major online grocer, faces this problem daily in its automated fulfillment centers.

Real-world examples

• 1.Automated warehouse picking robots
• 2.Robots handling varied inventory items in logistics centers
• 3.Robotic arms used for bin picking where items are loosely arranged
• 4.Ocado's automated grocery fulfillment centers