GENPACK: KPI-Guided Multi-Objective Genetic Algorithm for Industrial 3D Bin Packing

TL;DR

The paper tackles industrial 3D bin packing by introducing a KPI-driven hybrid genetic algorithm (Hybrid-GA) that integrates multiple industrial KPIs into a single fitness objective. It employs a layer-based chromosome and a three-phase pipeline (constructive heuristics, GA refinement, and post-processing) to produce stable, dense, and balanced packings, validated on 1,500 BED-BPP orders. The approach yields up to $35\%$ higher space utilization and $15$–$20\%$ stronger surface support with lower variance, at a modest runtime cost that remains feasible for batch deployment. These results demonstrate that explicit multi-KPI optimization can deliver robust, deployable packings in real-world palletizing scenarios, addressing core industrial constraints beyond mere utilization. Future work includes adaptive KPI scalarization, online/uncertain settings, and tighter integration with robotics and feasibility checks.

Abstract

The three-dimensional bin packing problem (3D-BPP) is a longstanding challenge in operations research and logistics. Classical heuristics and constructive methods can generate packings quickly, but often fail to address industrial constraints such as stability, balance, and handling feasibility. Metaheuristics such as genetic algorithms (GAs) provide flexibility and the ability to optimize across multiple objectives; however, pure GA approaches frequently struggle with efficiency, parameter sensitivity, and scalability to industrial order sizes. This gap is especially evident when scaling to real-world pallet dimensions, where even state-of-the-art algorithms often fail to achieve robust, deployable solutions. We propose a KPI-driven GA-based pipeline for industrial 3D-BPP that integrates key performance indicators directly into a multi-objective fitness function. The methodology combines a layer-based chromosome representation with domain-specific operators and constructive heuristics to balance efficiency and feasibility. On the BED-BPP benchmark of 1,500 real-world orders, our Hybrid-GA pipeline consistently outperforms heuristic- and learning-based state-of-the-art methods, achieving up to 35% higher space utilization and 15 to 20% stronger surface support, with lower variance across orders. These improvements come at a modest runtime cost but remain feasible for batch-scale deployment, yielding stable, balanced, and space-efficient packings.

Figures (3)

• Figure 1: Three-phase pipeline architecture: heuristics generate initial placements, a genetic algorithm refines residuals, and post-processing optimizes the final configuration.
• Figure 2: Qualitative (left) and quantitative (right) comparison of packing performance across algorithms on two representative orders from the BED-BPP dataset kagerer2023bedbpp. The left panel illustrates structural differences in packing outcomes, while the right plots absolute density and surface support as functions of packed items, with order-size distribution shown below.
• Figure 3: Supplementary analysis of additional KPIs, including height-width ratio, relative density, side support, and center-of-gravity balance as functions of packed items, with order-size distribution shown below.