DPAL solves the problem of localizing the top-layer of a pallet stack. It integrates with the path-planning system of a commercial robot arm. All of this is done in real-time, minimizing stoppages and keeping the robot in motion. Our solution uses an ifm O3D sensor mounted in a fixed location above the operations area. Using the 3D data gathered from the sensor, we provide a 4 degree-of-freedom pose (x, y, z, yaw) of the top layer to the robot arm control system. We also provide a confidence metric that can be used in the decision-making process. This provides visual feedback to the robot arm path planner to complete the pick. Layer imaging occurs in parallel with robot arm motion to minimize pick latency. This is essential for maintaining productivity in the application, especially when moving large loads with high inertia.
The example below shows a "typical" pick. The layer being imaged has well defined edges and a well-defined top-surface. This would produce a high confidence pick.
The next graphic shows the data from a pick where the surface of the layer is poorly defined. The perception challenges include: signal attenuation due to plastic wrap, non-uniform edges, and an irregular layer surface. Despite these challenges, the layer is accurately segmented but with a lower confidence score.
A common problem in full layer depalletization is that of an orphan case. In the absence of 3D perception, this will cause a costly tool crash. Our DPAL system reliably detects orphan cases, flags them as obstacles, and communicates this information to the robot controller.
Love Park works with our clients to identify the use-cases needed for successful operation of their full-layer depalletizing (or palletizing) applications. We then use these to implement the robust, high-performance application that you need for the highest levels of productivity.
Please contact us to discuss your full-layer depalletization or palletization needs.