Estimated Time of Arrival (ETA)

Reliable ETAs (Estimated Times of Arrival) form the basis for planning in logistics. Almost 60% of all supply chain managers criticize their current ETA solutions & predictions. Calculating the ETA (Estimated Times of Arrival) and thus the travel duration is one of the most difficult areas in planning in road freight transport, transport logistics and passenger travel especially for long routes and trucks.

We at Swarm Logistics have developed a novel approach for this. And we are extremely proud of it. To do this, we use a new proprietary artificial intelligence model, which is queried directly by our routing engine and which learns based on historical traffic flows and can very precisely predict the travel time of cars, vans and trucks; especially important, if you don’t plan for the moment, i.e. Arrival time is less than 1 hour, but for a point in time in the future, e.g. for freight rates, appointments for ramps, or over long distances with trips over multiple days, etc. Even planned a few hours in advance, “live traffic” based systems are no longer sufficient. (e.g. live traffic routing, live speedometer data).

Product Features

AI based ETAs prediction for cars, vans & trucks in Europe. Travel time estimation with a focus on planning.

By better predicting travel times, routes and tours can be planned much more realistically and driving and rest times and delivery promises can be adhered to. Costs can be better estimated and ramp times or other dates such as opening times can be adhered to. The AI has already been trained with millions of data points and continues to learn continuously. Tests have shown that our AI is already so good that even for the point in time “now” there is less than 3% deviation compared to the world’s leading routing engines with live traffic, i.e. e.g. on 6h travel time 3min and ETAs determined for the future ( i.e. everything from 2 hours) is far superior, especially for trucks. We conducted multiple pilot projects analyzing with ober 6000 routes for trucks, that the ETA is just deviating between 3-10% of real driven routes. The calculation speed for a route e.g. Lisbon to Moscow is approx. 900ms. The current model LUT 4.8 is trained on billions of data points and LUT 5 is coming soon and will be trained on over 12 times the amount of data.

Use the ETA for early notification

• Planning ramps/ramp slots
• Cross docking
• Tour planning etc.

Integrate the ETA service into your system via REST API, you send a start and end geolocation or address including a vehicle profile (car, van, truck) and receive an estimated travel time, distance and (if cost factors were provided) too, also a cost estimate. The service time of stops and possible breaks and refueling stops must still be added up. We can consider multiple breaks on the route (after a specific time).

For an “AI custom development” for your use case, in which the other parameters should also be predicted, e.g. service time, you are welcome to contact us.