In today’s logistics landscape, efficiency is a competitive necessity. Route optimization drives performance in everything from parcel delivery to complex warehouse fulfillment. Yet many companies still depend on outdated or rigid solvers that limit flexibility and performance.
What if your routing engine could deliver smarter solutions, faster?
TQrouting sets a new benchmark for high-performance route optimization. Built for both speed and solution quality, it delivers superior results across diverse vehicle routing problems (VRPs), enabling logistics teams to achieve more with fewer resources.
Why Another VRP Solver?
While there are plenty of VRP solvers available, many fail to scale or handle real-world complexity. TQrouting was built to close this gap, combining advanced optimization with machine learning techniques in an intelligent architecture that solves complex VRPs faster and more reliably.
From classic Capacitated Vehicle Routing Problems (CVRPs) to highly customized scenarios with client-specific constraints, TQrouting has a clear goal: deliver optimal solutions, faster.
Key Features & Customization
TQrouting supports problem-specific requirements, making it adaptable across industries and applications. Examples of its versatility include:
- Load capacity constraints: Manage varying loads and truck capacities efficiently.
- Time windows: Ensure clients are visited within required delivery windows.
- Service durations & route limits: Consider customer-specific service times and respect regulated driving and working time restrictions.
- Profit-based routing (Prize-collecting VRP): Maximize collected profit when serving all customers isn’t possible due to cost or resource limits.
- Driver shifts: Incorporate varying driver availability and working hours.
- Flexible start and end locations: Define different start and end locations or enable multiple depots setting and dynamic re-routing.
- Heterogeneous fleets: Optimize routing for fleets with diverse vehicle types and characteristics.
Beyond these, TQrouting’s modular architecture enables integration of new constraints, making it capable of adapting to a wide range of industry-specific requirements.
TQrouting’s modular design allows easy customization to client-specific needs. For example, it has delivered excellent performance in an intra-logistics routing problem for order fulfillment in a high-rack Very-Narrow-Aisle (VNA) warehouse. Unlike textbook VRPs, this scenario required accounting for aisle blockages caused by multiple VNA trucks to prevent collisions, while also managing real-time changes through dynamic planning.
Benchmark Results
We evaluated TQrouting’s performance against standard datasets for various VRP types with different constraints and objectives, measuring solution quality within short runtimes. We computed the relative gaps[1] to best-known solutions (BKS) found by state-of-the-art algorithms from research, typically designed for those specific problem types and solved on specific hardware with higher runtimes.
- CVRP (Uchoa et al., up to 1000 customers): 0.46% average gap to best-known solutions (within 1 min runtime).
- CVRP with Time Windows (Solomon, Gehring & Homberger, up to 1000 customers): 1.68% gap (within 1 min runtime).
- Multi-Depot VRP (Cordeau et al., up to 360 customers): 0.02% gap (within 1 min runtime).
- Team Orienteering Problem (Chao et al., Dang et al., up to 400 customers): 0.02% gap (within 1 min runtime).
- Large CVRP (Arnold et al., up to 30,000 customers): 3.00% gap with a 10 min runtime limit, 2.50% gap within 60 min runtime.
In practice, these results demonstrate trustworthy, high-quality routing decisions at operational speed. TQrouting can consistently find better solutions than other solvers within the same time limit, and/or reach the same solution quality faster than its competitors. For industries with large delivery networks, this translates into potential for millions in annual savings through reduced fuel consumption, lower personnel costs, and more on-time deliveries. TQrouting ensures that routes are efficient, reliable, and fully scalable.
Stay tuned for upcoming articles with detailed comparisons between TQrouting and other commercial and open-source solvers.
[1] All the experiments were run on a Google Cloud Platform’s virtual machine with 16 vCPUs and 32GB of memory.
A Path Into The Future of Routing
While most solvers are still catching up with modern logistics challenges, TQrouting is built for where the industry is heading, enabling logistics providers to achieve next-level efficiency.
Our research team continues exploring advanced acceleration via distributed and parallel computing, on CPUs, GPUs, and QPUs to improve large-scale VRP performance in shorter runtimes. This work positions TQrouting to seamlessly leverage quantum resources as hardware and software ecosystems mature—offering users a future-proof edge.