Robotic 3D Packing Planning for Mixed-Object Container Loading
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automated packing systems struggle to handle a wide range of objects reliably for commercial use, often resulting in inefficient container size selection, high shipping costs, and potential damage due to unforeseen errors during the packing process.
Innovation Solution
A robotic system with object detection sensors and a controller generates three-dimensional models of objects, plans a packing sequence, and corrects errors dynamically using closed-loop vision and manipulation to ensure stable and efficient packing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automated loading systems are used for mixed palletizing of boxes, then productivity is improved, but reliability deteriorates because they cannot handle a wide range of objects reliably
Solution Approach 1:
The system dynamically adjusts packing parameters such as container selection, object orientation, and placement position based on real-time vision system feedback. The controller modifies the packing plan parameters when deviations are detected, allowing the system to maintain high throughput while adapting to various object types and unexpected conditions during packing
Solution Approach 2:
The vision system continuously monitors the packing process and provides feedback to the controller. When objects are misplaced or packing deviations occur, the system detects these errors in real-time and adjusts subsequent packing actions accordingly, ensuring reliable handling of diverse objects while maintaining high productivity
2Adaptability or versatility
If human packers perform package fulfillment, then adaptability is improved, but productivity deteriorates due to cognitive taxation and slower pace
Solution Approach 1:
The system replaces human cognitive decision-making with an automated controller that uses vision system data to determine packing plans. The controller automatically selects containers, determines object orientations, and directs the robotic arm based on object recognition and packing optimization algorithms, eliminating cognitive taxation while maintaining adaptability to various object types
Solution Approach 2:
The system autonomously handles the entire packing process without human intervention. The vision system automatically identifies objects, the controller generates packing plans, and the robotic arm executes placements. The system self-corrects errors by detecting deviations and adjusting subsequent actions, achieving both high speed and adaptability
3Reliability
If conservative planning with geometric margins is used, then reliability is improved, but volume utilization deteriorates
Solution Approach 1:
The system uses dynamic margin adjustment rather than fixed geometric margins. The vision system continuously monitors actual object positions, and the controller adjusts placement margins in real-time based on detected deviations. This allows the system to maintain reliability by adding margins only when deviations are detected, rather than consistently reducing volume utilization
Solution Approach 2:
The vision system provides feedback on actual object positions and packing deviations. The controller uses this feedback to dynamically adjust placement strategies, adding geometric margins only when necessary to correct deviations or prevent collisions, thereby maintaining reliability while maximizing container volume utilization
Data Source
AI summary
Embodiments described herein relate to a system for packing objects in a container. The system may include a robotic device with a robotic arm, a plurality of object detection sensors, and a controller including at least one processor and a non-transitory computer-readable medium. The non-transitory computer-readable medium may store a set of program instructions, and the at least one processor may execute the program instructions including the operations of sensing a measurement of each object among a plurality of objects with at least one object detection sensor of the plurality of object detection sensors, obtaining a three-dimensional model for each object, determining a packing plan for the plurality of objects based on the three-dimensional model for each object, and loading, by a robotic arm, at least a portion of the plurality of objects into a container according to the packing plan for the plurality of objects.