Virtual Packaging Recommendation System

Project Brief: Virtual Packaging Recommendation System with Niryo Cobot Integration
Project TitleVirtual Packaging Recommendation System with Niryo Cobot Integration for Odd-Shaped Products in Amazon Fulfillment Centers
ObjectiveThis project aims to enhance Amazon’s fulfillment process by developing a virtual recommendation system integrated with the Niryo Cobot. The system calculates optimal packing layouts for up to 10 irregularly shaped products, visualizes these layouts, and automates packing via cobot execution. This solution is expected to reduce wasted space, improve container utilization, and accelerate packing times.
Problem StatementPacking products with irregular shapes efficiently within containers is challenging due to volume constraints and orientation requirements. This project seeks to create a system that generates ideal packing layouts for up to 10 odd-shaped products, reducing wasted space and minimizing shipping costs. Automation with the Niryo Cobot will facilitate precise implementation of optimized layouts.
Scope
Inputs: Up to 10 odd-shaped products with unique dimensions.
Outputs: A visual layout and automated execution of the optimal packing configuration, displayed on a screen for workers and implemented by the Niryo Cobot.
Features:
Calculation of product volumes and orientations.
Mathematical modeling for packing optimization.
Real-time visualization and cobot-assisted execution.
Interface for fulfillment personnel to monitor and interact with the layout.
Project PhasesThe project consists of three main phases:
Math Model and Shape Creation: Develop a mathematical model and generate irregular shapes from paste, creating the foundation for packing layouts (completed).
Virtual Enclosures: Create virtual enclosures based on the mathematical model results for the odd shapes. These enclosures represent the virtual space each object occupies, providing an alternative packaging solution.
Automation and Comparison: Program the Niryo Ned One robot to execute the packing layouts. Compare the packaging efficiency of layouts optimized for linear objects with those optimized for virtual enclosures, aiming to determine which approach maximizes packaging space utilization.
Key Deliverables
Mathematical Model: Calculates the volumes and orientations of both linear and enclosed shapes.
Virtual Enclosure Algorithm: Generates virtual enclosures for each irregular object, incorporating them into the model to compare packaging efficiencies.
Cobot-Control System: Commands the Niryo Cobot to execute the model’s recommended orientations and placements.
Visualization Interface: Displays the recommended layout and monitors cobot activities.
Testing Scenarios: A series of test cases with varied product shapes and sizes to validate system precision and layout efficiency for both linear and enclosed configurations.
Mathematical Model
The mathematical model’s goal is to determine the smallest possible dimensions aa, bb, and cc for a container that can accommodate all 10 rectangular objects without overlap while minimizing wasted space.
Objective Function:Minimize dd such that d≥a,d≥b,d≥cd≥a,d≥b,d≥c.
Constraints:
Box Dimension Constraints: Ensures that the container’s dimensions meet the required bounding box:
d≥a,d≥b,d≥cd≥a,d≥b,d≥c
Non-Overlap Constraints: To avoid overlapping between rectangles RiRi and RjRj, at least one of the following must hold:
∣xi−xj∣≥li+lj2,or∣yi−yj∣≥wi+wj2,or∣zi−zj∣≥hi+hj2∣xi−xj∣≥2li+lj,or∣yi−yj∣≥2wi+wj,or∣zi−zj∣≥2hi+hj
These are reformulated using binary variables uij,kuij,k to ensure that at least one direction has non-overlapping conditions satisfied.
Containment Constraints: Each rectangle must remain within the container bounds:
0≤xi≤a−li2,0≤yi≤b−wi2,0≤zi≤c−hi20≤xi≤a−2li,0≤yi≤b−2wi,0≤zi≤c−2hi
This model integrates non-overlapping and containment constraints with an objective function minimizing the smallest bounding dimension, dd, to maximize space utilization.
Solution for Linear Shapes
The current mathematical model has yielded an optimal packaging layout for linear objects, achieving a smallest bounding dimension (d) of 10.0. Here are the placements of each rectangle:
R1: Center (x=1.15, y=2.55, z=6.1), Dimensions: 2.3 x 5.1 x 2.6
R2: Center (x=2.015, y=3.15, z=3.405), Dimensions: 4.03 x 6.3 x 2.41
R3: Center (x=2.55, y=8.15, z=3.35), Dimensions: 5.1 x 3.7 x 2.3
R4: Center (x=2.35, y=2.25, z=1.1), Dimensions: 4.7 x 4.5 x 2.2
R5: Center (x=6.93, y=1.55, z=3.5), Dimensions: 5.8 x 3.1 x 2.6
R6: Center (x=6.8, y=1.9, z=1.05), Dimensions: 4.2 x 3.8 x 2.1
R7: Center (x=6.33, y=4.375, z=3.5), Dimensions: 4.6 x 2.55 x 2.6
R8: Center (x=7.25, y=7.55, z=1.6), Dimensions: 4.3 x 3.8 x 3.2
R9: Center (x=2.45, y=6.55, z=6.1), Dimensions: 4.9 x 2.9 x 2.6
R10: Center (x=2.85, y=1.45, z=8.7), Dimensions: 5.7 x 2.9 x 2.6.
This layout provides a baseline for comparing future layouts that use virtual enclosures, aiming to identify the configuration that maximizes packaging space utilization.
Methodology
Data Collection: Gather product dimensions and analyze existing packing patterns.
Modeling: Use convex hulls and phi-functions to model shapes and virtual enclosures, employing linear programming for layout optimization.
System Development: Develop algorithms for both linear and enclosed layouts and integrate cobot control.
Testing: Conduct efficiency tests, comparing space utilization between linear and enclosure-based layouts to find the optimal approach.
Resources Required
Personnel: Mathematicians, data scientists, software engineers, robotics specialists, UX/UI designers.
Hardware: Niryo Cobot, visualization hardware.
Software: 3D modeling tools, optimization libraries (e.g., Gurobi, SciPy), visualization tools (e.g., Unity, WebGL).
Challenges
Modeling complex geometries and virtual enclosures.
Ensuring cobot adaptability to various shapes.
Developing an interface to support both cobot control and worker interaction.
Success Metrics
Packing Efficiency: Optimal space utilization for both linear and enclosure-based layouts.
Time Efficiency: Reduced packing times due to cobot automation.
User Satisfaction: Positive feedback on usability and layout accuracy.
ConclusionIntegrating the Niryo Cobot into this Virtual Packaging Recommendation System will provide Amazon with an efficient, automated packing solution that optimizes space use for odd-shaped products. This system’s comparative approach will guide the choice of linear versus virtual enclosure layouts, ultimately improving container utilization, reducing costs, and streamlining fulfillment operations.
 ... Show more