GPU-Accelerated Truck Planning

NVIDIA cuOpt system for real-time intermodal transport optimization

📅 September 2025 - January 2026
🏢 Move Intermodal
🔧 GPU Computing, VRP
📍 Week 13-14 (Refinement Phase)

⚠️ This project is under NDA with Move Intermodal. Specific company data, routes, and client information cannot be publicly shared.

🔄 This project is currently in the refinement phase (Week 13-14). The information below describes the current status and remaining work.

Project Overview

An 18-week academic research project for Move Intermodal developing a proof of concept for a GPU-accelerated optimization system using NVIDIA cuOpt for automated truck planning. The system supports planners in optimizing the current manual planning process by matching orders with trucks based on formally documented business rules. The data pipeline is integrated with Snowflake and processes orders, fleet status, and driver availability to generate optimal routes with minimal empty kilometers and maximum fleet productivity. The end result is an interactive dashboard where planners can visualize, evaluate, and manually adjust optimized truck schedules as needed.

Challenges

The current manual planning process results in suboptimal routes, excessive empty kilometers, and inefficient truck capacity utilization. Planners must manually match trucks to jobs, calculate driver start times, verify delivery windows, and use geographical knowledge for container reload operations. The core problem is an NP-hard multi-constraint optimization challenge where all business rules (ADR certification, time windows, container availability, driver preferences) must be balanced simultaneously. The biggest challenge is translating partially undocumented planning rules into an automated model that reaches optimal solutions within reasonable time.

Progress & Expected Results

The project started in September 2025 and runs until January 2026 (18 weeks). Infrastructure is fully set up with GPU compute environment and Snowflake connectivity. The ETL pipeline processes orders, trucks, and driver data with data quality checks. Currently (week 13-14) the project is in the refinement phase with focus on fine-tuning cost functions, constraint weights, and algorithm parameters. The first MVP (Working Planning Tool) was delivered on 28/11/2025, the fully optimized system will be completed on 22/01/2026. Expected impact: 15-20% reduction in kilometers driven, cost savings per kilometer through optimized routes, and reduction of planning time from 2-3 hours to <5 minutes.

18
Week Project
60
Trucks Fleet
GPU
cuOpt Acceleration
Week 13-14
Current Status

Why GPU Optimization?

Benefits of GPU acceleration for route optimization:

Complex Constraint Management

The system processes three main constraint categories that were implemented incrementally (week 7-12). Each constraint has hard variants (must comply) and soft preferences (optimization goals). Implementation followed a phased approach: first order-specific restrictions (week 8), then truck and driver requirements (week 8), followed by full optimization (week 9-10), and finally complex container restrictions (week 11-12):

📦 Order-Specific Restrictions

  • ADR transport requirements (hazardous materials)
  • Specific container type needs
  • Chassis requirements and compatibility
  • Time slots at loading/unloading locations
  • Customer service-level agreements
  • Priority levels (urgent/normal)

🚛 Truck & Driver Restrictions

  • Transport type and modality
  • ADR certification drivers
  • Specific chassis connections
  • Maximum truck weight and capacity
  • Driving time limits and EU working hours
  • Driver preferences and weekend availability
  • Start location and ongoing orders

📍 Container & Location Restrictions

  • Container availability at order location
  • Load/unload sequence verification
  • Geographical knowledge for reload operations
  • Emission zone restrictions
  • Cross-docking requirements
  • Customs procedure time

🎯 Why is this technically challenging?

Combining all these constraints in one optimization model is an NP-hard problem. Traditional solvers use branch-and-bound or constraint programming, but these scale poorly with high complexity. GPU acceleration enables exploring much more solution space through parallel search in the same timeframe.

Technical Implementation

Technology Stack

🎮 NVIDIA cuOpt
🐍 Python
❄️ Snowflake
🗄️ SQL

Team & Methodology

The project is executed by a 5-person academic research team following the CRISP-ML methodology:

The iterative development approach combines weekly sprints with regular feedback from Move Intermodal planners. Each Function Plan step was documented and validated before moving to the next phase.

Optimization Pipeline

The system follows a phased pipeline based on the Function Plan (week 7-14):

Snowflake Data Integration

The system integrates with Snowflake for real-time data access. Each morning the pipeline is triggered for next-day planning. The data pipeline retrieves the following:

⚡ GPU Acceleration Advantage

The NVIDIA cuOpt API uses parallel computing on GPUs. Where a traditional CPU solver must sequentially evaluate thousands of route combinations, the GPU can do this in parallel on thousands of cores simultaneously. This reduces solution time from hours to less than 5 minutes, even with complex multi-constraint optimization.

Project Status & Roadmap

18
Weeks Total
✅ 28 Nov
MVP 1 Delivered
22 Jan
MVP 2 Deadline
Week 13-14
Current Phase

Development Phases (Function Plan)

Step 1: Basic Order & Truck Loading (Week 7)
Implementation of load_orders(), load_trucks(), create_planning() functions for basic truck-order matching on one specific day.
Step 1A: Restrictions Dataset Documentation (Week 7)
Parallel to step 1: documentation of all restrictions per dataset for future implementation. Feature Overview Inventory was built.
Step 2: Order-Specific Restrictions (Week 8)
Adding ADR requirements, container types, chassis needs, time slots, and customer priorities to orders dataset.
Step 3: Truck & Driver Restrictions (Week 8)
Implementation of transport type, ADR certification, chassis connections, weight limits, driving time limits, and start locations.
Step 4: Optimization with Constraints (Week 9-10)
Full implementation of constraints_validation() and cost_function(). cuOpt optimization with all restrictions except containers.
Step 5: Container Restrictions (Week 11-12)
Complex container location verification and load/unload sequence checking. Correction of incorrectly labeled data in datasets.
🔄
Step 6: Advanced Optimization & Refinement (Week 12-14)
Fine-tuning of cost functions, constraint weights, and algorithm parameters for maximum efficiency. Completion of Working Planning Tool MVP.

Expected Impact (Business Requirements)

Project Milestones & Deliverables

Technical Challenges & Innovations

Why is this project technically interesting?

  • Phased constraint implementation: Combining order, truck, and container restrictions forms an NP-hard optimization problem. By adding constraints incrementally (following the Function Plan from week 7–12), the process remained manageable and testable.
  • Real-time Snowflake integration: The data stream processes daily over 60 trucks with historical patterns, current locations, and order updates. The optimizer remains stable even with incomplete or inconsistent data.
  • Hybrid optimization: A mix of exact algorithms for hard constraints and heuristics for soft preferences. Thanks to GPU acceleration, both methods can be executed within a 60-second time window.
  • From implicit knowledge to formal logic: Planner knowledge such as geographical routes or driver preferences has been translated into explicit constraints and cost functions within the optimization model.
  • Research with production impact: An 18-week CRISP-ML trajectory with strict documentation and evaluation, focused on a real industrial challenge for a company with over €100 million revenue.

Check out my other projects