(only for Maps for Germany available)
Swarm routing is purpose build for commercial vehicles and fleets and is used as an underlying technology in our other solutions deriving tour-plans, tour optimization and to estimate costs for freight rates.
Route planning seems to be a problem that has been solved: drive from A to B. Value-based routing is a new approach. While the shortest distances or fastest routes can be planned for consumers, the situation is different in a commercial environment. It’s all about costs! Value-based routing uses cost factors, which can be different for each vehicle, driver, trailer, and their combinations, and evaluates routes; it calculates routes with minimal costs rather than minimizing distances. Two-man crews, 3.5t or 40t vehicles with toll are all accounted for differently and this results in completely different routes.
Calculating the estimated times of arrival (ETA) and thus the driving time is one of the most difficult areas in route planning. To do this, we use our proprietary artificial intelligence model.
Initially we didn’t address routing; we thought it was a solved problem and just use an existing routing solution and build on top of it. In the meantime, while developing and testing our tour-planning and optimization solutions, we realized, that the available routing solutions, where not sufficient for our purposes. The metrics, where not available to evaluate single freight or transportation orders properly, nor was the problem solved that routing is needed not only for execution and the time point now, but also for planning for any time point in the future, especially estimated times of arrival were far away from reliable for planning cases and thus not working for tour-planning and freight rate calculation. The tested truck routing solutions on the market were even more unreliable.
So, we forked an open-source routing solution and extended it over multiple years not to be one fits all, but purpose built for commercial vehicles and fleets. We believe now that our solution is the best-in-class solution.
What is new and what to test?
It is a new approach for routing: “Value-based Routing” is finding routes with minimum costs based any vehicle, driver, trailer specific cost factors, considering toll roads to drive or not to drive depending on what is more cost optimal in total instead of minimizing distances.
It is giving you immediately the route costs, on which you can derive your freight rates considering heterogeneous vehicles in size, weight, height and costs, and other different vehicle restrictions incl. no transit zones for trucks, but free for delivery and multiple on-board members etc.
We not only introduced this new form of routing, but furthermore how it is handling the time calculation is completely new. Swarm routing is using a Machine Learning approach to estimate travel time without using live traffic information.
It is not adding delays up, but rather the routing engine is triggering and processing the information while routing and thus give more accurate results. We also use an unconventional approach to calculate the times by not tracking devices.
We have seen in test that the machine learning derived driving times and ETAs (current in the model LUT Version 4) are deviating in time prediction from the leading routing solutions (that are using live traffic information and we just the ML predictions) for long routes under 2% (these are minutes deviations on multiple hours of driving) and short routes under 4%. And our tests showed that we had for trucks by far the most precise solution in comparison to the tested ones.
Furthermore, we modeled the problem differently in an unconventional way and were able to extrapolate it to achieved similar or in some cases much better results with much less data in comparison to other solutions that collect terabytes of data daily with in comparison to our 100Gb data in total. With our approach there is no need to track billions of devices to analyze the device movements.
As a side note the training of model LUT V5 is in progress, and it has more than double the amount of data than LUT V4, so we expect an increase in results quality.
Freight rates calculations, service task & ramp times appointments will become much more reliable for any time point in the future. With better times estimated, there will be less legal driving time violations for drivers, the tours can be clocked better and there will be less need for time safety buffers e.g., not to lose your loading ramp slot or miss a customer appointment.
Feel free to test it and we appreciate your feedback on the results.
Developers can now integrate Value-based Routing directly into their apps and products through our API. Also prepare for our cost calculation service, which uses over 250 variables directly out of your ERP system and is feeding the Value-based Routing engine instead of the here shown few variables.
Stay tuned for the next things, the other solutions will also be available as demos in a short time.
Please spread the word and share the demo link. Thank you being with us on the journey and coming closer to a self-organizing supply chain as part of the machine economy!
We are looking for pilot partners to further validate the products and in development solutions in the field.
Questions and feedback please to email@example.com
Thank you and enjoy the demo.