Let's load your experience
To make your visit as smooth as our service, we use cookies. Ready to load an optimized journey
EVA
The artificial intelligence that plans, verifies and coordinates your container logistics
Request a demo
Context
Worksitelogisticsstillrelyonphonecallsandmanualplanning.EVA(EngineforVirtualAnticipation)automatestheseprocessesandbringscooperativeintelligencetotheheartofoperations.
Challenges to address
Nereva simplifies transport logistics and tackles the challenges of managing transport requests
Plan transport optimally
A transport company must allocate its requests each day among trucks with different capacities, aiming to complete as many missions as possible while minimizing downtime.This requires considering task compatibility with each vehicle, dependencies between successive requests (such as warehouse stops), and the frequent uncertainties on worksites, delays, cancellations, changes in priority which require quick rescheduling.The goal is to maximize the number or value of completed missions while reducing costs, ensuring flexibility and real-time responsiveness, something no static planning system can achieve today.
Automatic verification
When waste is collected, it is sent to a sorting center whose processing capacity and costs vary. Precise identification of waste is therefore essential to avoid environmental, billing, or contractual errors.In Nereva, drivers must photograph each step of their service, but there is no guarantee that these images are always relevant. This is why an automatic photo recognition system is necessary: it would validate the evidence, identify the types of waste in the containers, and even optimize routes by directing transporters to the most suitable sorting center.The goal is to develop these models from the field photos collected via Nereva.
Creating a dataset
Thanks to the transporters’ mobile app, it will be possible to collect a wide range of data: routes, speeds, requests, unexpected events, delays, and much more.By ensuring the anonymization of this data, it can be made available for different projects, whether academic or private.
Explainability of results
Today, logistics managers maintain full control over their company’s planning. When an intelligent system proposes solutions, it is essential that its results are understandable and acceptable to them.Each decision must be explainable, and its benefits clearly visible, to avoid any loss of trust or obstacles in adopting the system.
Expected results
Every photo taken is automatically validated as evidence
An open dataset gathering all information
A clear and understandable user experience
Toachievetheseresults,EVAdoesnotrelyonasinglealgorithm,butonseveralspecializedintelligentagentsworkingtogether;eachdedicatedtoaspecifictasktoensuregreateraccuracyandflexibility.
Optimization agents
Optimization agents are responsible for organizing transport schedules. Identical and cooperative, they explore different combinations to allocate requests among trucks in the most efficient way possible, minimizing empty trips and adapting to on-site disruptions.
Detection agents
Detection agents use models capable of identifying specific elements, such as a skip or a type of waste in a photo. They can be integrated directly into the mobile application to perform real-time detection and ensure the quality of evidence collected in the field.
Coordination agents
Coordination agents ensure communication between the other agents. They receive new information, route it to the appropriate agents, and ensure that each piece of data is used at the right time to optimize system decisions.
Theorchestratoragentisconnectedtothedetectionagentsaswellastotheoptimizationsubspaces.Anoptimizationsubspacerepresentsaportionoftheoverallproblem,definedsothatchangesmadewithinitdonotaffecttheothersubsets,butonlythefinaloutcome.Basedontheinformationreceived,theorchestratorroutesdatatothemostappropriateagentsandthenstorestheresultsinthedatabasesotheycanbeleveragedbyNereva.
Request a demo
Technology and research
From prototype to autonomous intelligence
EVA evolves step by step, moving from research to real-world application. Each year marks a key milestone toward a more intelligent, more autonomous system, fully integrated into transport and construction operations.
2025
Foundations
Implementation of the architecture and first intelligent agents.
Centralization of field data collected through Nereva.
Initial prototypes validated on real-world use cases.
2026
Field intelligence
Training detection agents to identify waste types.
Integration of models into the mobile application.
Optimization tests conducted with logistics partners.
2027
Cooperation and integration
Agent collaboration to enrich and consolidate data.
Incorporation of new variables such as weather conditions.
Deployment of EVA in real environments as an assistant.
2028
Autonomy and explainability
Agents capable of adjusting schedules in real time.
Predictive models to anticipate demand.
Explainable and traceable decisions to build trust.
Join the EVA project
Request a demo