3d-bin-packing

What it does

The skill is an expert in 3D bin packing and three-dimensional space optimization. It aims to help pack 3D boxes and items into containers, trucks, or bins while maximizing space utilization, minimizing number of containers, and ensuring physical stability.

How it works

Initial Assessment

The agent asks for problem objectives (e.g., minimize number of containers, maximize container utilization, minimize shipping costs, ensure load stability), item specifications (count, dimensions, weights, fragility, stackability, rotation allowances), container specifications (dimensions, weight capacity, multiple types, quantity), physical constraints (weight distribution, center of gravity, stacking limits, fragility, load sequence), and any special requirements (multi-drop deliveries, grouping, axle limits, door access).

3D Bin Packing Framework

The problem is categorized into: 3D Bin Packing Problem (minimize bins, all items packed, no overlap, axis-aligned or with rotation); Single Container Loading Problem; Pallet/Container Loading Problem; Multi-Container Heterogeneous Packing; Weight-Constrained 3D Packing.

Mathematical Formulation

The model defines decision variables for item coordinates, bin assignment, orientation, and which bins are used. Objective is to minimize total bins. Constraints cover non-overlap, containment, support, weight, and stability; complexity notes NP-hard and reliance on heuristics.

Algorithms and Solution Methods

Exact Methods

An Integer Programming formulation is provided (solve_3d_bin_packing_ip) using Python pseudocode with libraries (e.g., pulp). It defines item positions, bin assignment, orientation, and bin usage; enforces assignment, containment, weight, and bin usage constraints. Non-overlap is noted as omitted in this simplified version.

Heuristic Methods

Bottom-Left-Back (BLB) Algorithm

Greedy placement at lowest, leftmost, back-most positions; includes 3D Box and Space representations, orientation handling, and a Space class with fits checks and rotation support. It places items in existing containers or creates new containers as needed, updating spaces and ensuring total weight does not exceed max_weight.

Layer Building Algorithm

Builds horizontal layers, packs items within each layer (2D problem), and stacks layers vertically. Includes a Layer Packing routine that uses packing within a layer and assigns 3D positions with a layer_z offset.

When to use it

Use when the user mentions 3D packing, container packing, box packing, 3D bin packing problem, cube packing, three-dimensional packing, cargo loading, or space optimization. For pallet-specific loading or container logistics, related references are provided but not required by this skill unless the prompt mentions those specifics.

What it can touch

The skill declares claude-code as a declared tool. It includes code blocks and Python-like snippets for both exact and heuristic methods, including function solve_3d_bin_packing_ip, Box and Space classes, bottom_left_back_packing, create_residual_spaces, calculate_3d_utilization, and layer_building_algorithm. These references indicate it can touch code-based solutions and 3D packing logic.

Caveats

• The exact method notes that non-overlap constraints are omitted in the simplified version of the IP model.
• The provided snippets rely on external libraries (e.g., PuLP) and are described as practical for very small problems (<10 items).
• Rotation handling is described but may be context-dependent based on orientation definitions provided in the code.
• The license is MIT as per the skill metadata.