AI-industrie.png

Smart Manufacturing & Materials-iStock-1135159658-BEWERKT.jpg

In Europe, 20% of trucks travel empty. The Achterhoek-based IT startup Bullit Digital wondered how that percentage could be reduced using artificial intelligence.<img src="https://cms.aihub-oost.nl/uploads/w_JQBW_6_JVOD_Rs_P0_J5_R_Jax_Ok_Jo_Wt_Oa_H_Ocshdx1j6_X1_9f76a01591.jpeg" alt="wJQBW6JVODRsP0J5RJaxOkJoWtOaHOcshdx1j6X1.jpeg" srcset="https://cms.aihub-oost.nl/uploads/thumbnail_w_JQBW_6_JVOD_Rs_P0_J5_R_Jax_Ok_Jo_Wt_Oa_H_Ocshdx1j6_X1_9f76a01591.jpeg 234w,https://cms.aihub-oost.nl/uploads/small_w_JQBW_6_JVOD_Rs_P0_J5_R_Jax_Ok_Jo_Wt_Oa_H_Ocshdx1j6_X1_9f76a01591.jpeg 500w,https://cms.aihub-oost.nl/uploads/medium_w_JQBW_6_JVOD_Rs_P0_J5_R_Jax_Ok_Jo_Wt_Oa_H_Ocshdx1j6_X1_9f76a01591.jpeg 750w,https://cms.aihub-oost.nl/uploads/large_w_JQBW_6_JVOD_Rs_P0_J5_R_Jax_Ok_Jo_Wt_Oa_H_Ocshdx1j6_X1_9f76a01591.jpeg 1000w," sizes="100vw" width="1024" height="682">By linking trips together via an intelligent agent, they can be completed more efficiently and, as a result, more sustainably. Bullit developed a proof-of-concept for Emons Group, a niche player in the European transportation market that focuses on innovation and sustainable transportation solutions.<h2>Dataset</h2>A dataset was used containing a list of trips that carriers can bid on. The dataset includes approximately 1,000 routes, for which information such as the pickup location, drop-off location, annual frequency, and distance in kilometers is known.Because the number of possible routes for the agent increases exponentially with each additional location, we chose to cluster the locations. By combining this with data on the number of trips departing or arriving, we can identify initial clusters of pickup and drop-off points where the vehicle is likely to travel empty.<h2>Reinforcement Learning</h2>For the proof-of-concept, reinforcement learning was used—a machine learning technique that allows an agent to learn based on a simulation. During training, the agent starts at a random pickup location each time, takes actions, and receives scores for those actions. Based on those scores, the agent continuously adjusts its neural network so that it can make better predictions in the future. Bullit Digital explains the approach step by step on its website.In just ten minutes of training, a negative score was turned into a positive one. After about 4,000 iterations, the system was still learning something new; the average score continued to rise slowly.Want to learn more about the possibilities of AI for logistics? Contact us via our contact form.

Making Smarter Choices in Logistics

These entrepreneurs have paved the way for you

Meet our frontrunners

wJQBW6JVODRsP0J5RJaxOkJoWtOaHOcshdx1j6X1.jpeg

In Europe, 20% of trucks drive empty. The Achterhoek-based IT startup Bullit Digital wondered how that percentage could be reduced using artificial intelligence.