Periodic Reporting for period 2 - LOGISTAR (Enhanced data management techniques for real time logistics planning and scheduling)
Período documentado: 2019-12-01 hasta 2021-09-30
The main objective of the LOGISTAR project was to allow an effective planning and optimization of transport operations in the supply chain by taking advantage of horizontal collaboration and relying on real-time data gathered from an interconnected environment. For this, real-time decision-making support tools and visualization tools of freight transport were developed. Their purpose was to deliver information and services to the various agents involved in the logistic chain.
The European Union faces the challenge of maintaining and increasing its economic growth and coping with an increasing freight transport demand and limited transport infrastructure in the next years and decades. Considering the crucial importance of the freight industry and its influence, there is a need to increase its efficiency.
LOGISTAR leveraged the data available related to logistics and transport, such as data coming from intermodal mobility transport (logistic infrastructures, transport schedules, prices, traffic congestion, accidents and road networks), to process it in real-time and to deliver services-based optimization and planning of resources and routes. In particular, this was done taking into account concepts such as synchromodality, prediction of disruptions in real-time, and collaboration through negotiation and identification.
The project was deployed in three actual Living Labs plus a Virtual one, to test its effectiveness in real operation environments and under more extreme circumstances.
LOGISTAR seeked to improve the logistic infrastructure by means of the use of cutting-edge ICT technologies. It aimed to reduce the distribution cost by 5 to 10%, loading factors (up to 10%) and to enhance the synchromodality, as the use of different transport means.
The European Union faces the challenge of maintaining and increasing its economic growth and coping with an increasing freight transport demand and limited transport infrastructure in the next years and decades. Considering the crucial importance of the freight industry and its influence, there is a need to increase its efficiency.
LOGISTAR leveraged the data available related to logistics and transport, such as data coming from intermodal mobility transport (logistic infrastructures, transport schedules, prices, traffic congestion, accidents and road networks), to process it in real-time and to deliver services-based optimization and planning of resources and routes. In particular, this was done taking into account concepts such as synchromodality, prediction of disruptions in real-time, and collaboration through negotiation and identification.
The project was deployed in three actual Living Labs plus a Virtual one, to test its effectiveness in real operation environments and under more extreme circumstances.
LOGISTAR seeked to improve the logistic infrastructure by means of the use of cutting-edge ICT technologies. It aimed to reduce the distribution cost by 5 to 10%, loading factors (up to 10%) and to enhance the synchromodality, as the use of different transport means.
The LOGISTAR project was launched in June 2018 and lasted 40 months. The first step in achieving the LOGISTAR objectives was to identify the end user needs and system requirements so that the LOGISTAR solution could solve the weaknesses in the current systems. For this, 22 interviews with companies from a range of industry sectors in five different countries were conducted and a coherent list of user needs and system requirements was produced.
Afterwards, the main research technologies of the project started to be developed:
- Data sources concerning the Living Labs were identified and specified and a LOGISTAR data model was defined. Also, the data acquisition pipelines, metadata, data enrichment and the semantic layer were implemented.
- A state-of-the-art analysis on artificial intelligence and optimization techniques applied to logistics has been performed.
- Algorithms for predicting timings and events for a logistics plan, and for learning and inference of stakeholder preferences for optimisation were developed and implemented.
- The global optimization routing engine, which includes the mathematical analysis of the model, as well as the design and implementation of the engine were achieved.
- An analysis of the negotiation strategies used by companies to promote co-loading and back-hauling were performed, as well as an implementation using electronic agents were delivered. This analysis and automatic negotiation includes routing re-optimization.
- The technical basis server infrastructure has been already designed and implemented.
- The Living labs for testing LOGISTAR solutions were setted up, as well as testing activities were carried out both at the level of particular systems and as global solutions.
The LOGISTAR consortium actively participated in external events to promote the results achieved with presentations in several conferences and through Users’ Board workshops. Partners also identified their exploitable results and are considering the best protection framework for the LOGISTAR platform.
Afterwards, the main research technologies of the project started to be developed:
- Data sources concerning the Living Labs were identified and specified and a LOGISTAR data model was defined. Also, the data acquisition pipelines, metadata, data enrichment and the semantic layer were implemented.
- A state-of-the-art analysis on artificial intelligence and optimization techniques applied to logistics has been performed.
- Algorithms for predicting timings and events for a logistics plan, and for learning and inference of stakeholder preferences for optimisation were developed and implemented.
- The global optimization routing engine, which includes the mathematical analysis of the model, as well as the design and implementation of the engine were achieved.
- An analysis of the negotiation strategies used by companies to promote co-loading and back-hauling were performed, as well as an implementation using electronic agents were delivered. This analysis and automatic negotiation includes routing re-optimization.
- The technical basis server infrastructure has been already designed and implemented.
- The Living labs for testing LOGISTAR solutions were setted up, as well as testing activities were carried out both at the level of particular systems and as global solutions.
The LOGISTAR consortium actively participated in external events to promote the results achieved with presentations in several conferences and through Users’ Board workshops. Partners also identified their exploitable results and are considering the best protection framework for the LOGISTAR platform.
HHere are the main advances beyond the state of the art:
Artificial intelligence focused on prediction
Predictive models were developed which can predict arrival times for different legs in logistics transportation, between specific locations, for specific vehicle types, at different hours of the day, days or the week, or times of the year. These were augmented with predictions of turnaround times at the destinations, which are again location, time and season specific. Finally, models for acquiring the preferences of specific companies and logistics staff in those companies for balancing the trade-offs between important objectives such as time, distance, cost and reliability of logistics components have been developed.
Global optimisation planning
Research on mathematical models and optimization meta-heuristics focused on the reduction of logistic costs in different real scenarios have been done. Based on this analysis, fouroptimization modules were deployed and integrated: twofor collaborative optimization, where 2 or more companies can share resources for better use and improvement of efficiency. A thirdone for multimodal optimization, considering the return of different solutions for selection by stakeholder, and a fourthone for the planning of the use of resources at the distribution center.
Automated negotiation and planning re-optimisation
Research on models of automated negotiation agents that propose the exchange of tasks. Every agent has a set of tasks (deliveries to be done) but can propose to other agents to perform part of these tasks on the basis of a pre-fixed compensation (according to the distance of the delivery). The agents can automatically suggest exchanges for those tasks that are expensive for them, detecting other agents for which they are cheaper, and who could accept them. This implies that every agent knows their own set of tasks, can optimize them, but also knows (part) of the task of other agents.
LOGISTAR delivered two services based on the advanced processing techniques:
• A control and decision-making tool for logistics operations capable of monitoring goods through the whole logistics procedure, allowing an integrated planning of resources and providing dynamic routing relying on synchromodality and horizontal collaboration among agents.
• Real time information of freight transport will be delivered by means of a website, where the position of the goods in real time in the various means of transport will be shown.
The project results contributed to enhance different aspects related to logistics and freight transportation. LOGISTAR improved the logistic infrastructure by means of the use of cutting-edge ICT technologies. These technologies were thoroughly tested in four different living labs and showed benefits in every single use case.
The LOGISTAR system calculated logistic improvements, showed collaborative cost savings, a decrease in empty kilometres and increase in vehicle fill rate. Additionally the emerging technologies used provided visibility on the logistic flow.
And while there are still difficulties in achieving these results from a practical standpoint, due to the fact that it is sometimes difficult to grasp a full human planning system by a single algorithm, there is a common understanding that the LOGISTAR platform can grow into a decision support tool for numerous use-cases in the transport and logistics sector.
Artificial intelligence focused on prediction
Predictive models were developed which can predict arrival times for different legs in logistics transportation, between specific locations, for specific vehicle types, at different hours of the day, days or the week, or times of the year. These were augmented with predictions of turnaround times at the destinations, which are again location, time and season specific. Finally, models for acquiring the preferences of specific companies and logistics staff in those companies for balancing the trade-offs between important objectives such as time, distance, cost and reliability of logistics components have been developed.
Global optimisation planning
Research on mathematical models and optimization meta-heuristics focused on the reduction of logistic costs in different real scenarios have been done. Based on this analysis, fouroptimization modules were deployed and integrated: twofor collaborative optimization, where 2 or more companies can share resources for better use and improvement of efficiency. A thirdone for multimodal optimization, considering the return of different solutions for selection by stakeholder, and a fourthone for the planning of the use of resources at the distribution center.
Automated negotiation and planning re-optimisation
Research on models of automated negotiation agents that propose the exchange of tasks. Every agent has a set of tasks (deliveries to be done) but can propose to other agents to perform part of these tasks on the basis of a pre-fixed compensation (according to the distance of the delivery). The agents can automatically suggest exchanges for those tasks that are expensive for them, detecting other agents for which they are cheaper, and who could accept them. This implies that every agent knows their own set of tasks, can optimize them, but also knows (part) of the task of other agents.
LOGISTAR delivered two services based on the advanced processing techniques:
• A control and decision-making tool for logistics operations capable of monitoring goods through the whole logistics procedure, allowing an integrated planning of resources and providing dynamic routing relying on synchromodality and horizontal collaboration among agents.
• Real time information of freight transport will be delivered by means of a website, where the position of the goods in real time in the various means of transport will be shown.
The project results contributed to enhance different aspects related to logistics and freight transportation. LOGISTAR improved the logistic infrastructure by means of the use of cutting-edge ICT technologies. These technologies were thoroughly tested in four different living labs and showed benefits in every single use case.
The LOGISTAR system calculated logistic improvements, showed collaborative cost savings, a decrease in empty kilometres and increase in vehicle fill rate. Additionally the emerging technologies used provided visibility on the logistic flow.
And while there are still difficulties in achieving these results from a practical standpoint, due to the fact that it is sometimes difficult to grasp a full human planning system by a single algorithm, there is a common understanding that the LOGISTAR platform can grow into a decision support tool for numerous use-cases in the transport and logistics sector.