Call: TCall number: RETECH – L1: Projectes per fomentar la transferència i innovació en l’àmbit de la IA.

SMARTCYCLE_IA transitions vacuum waste collection from static scheduling to a dynamic AI model based on patented URD algorithms. Using an Edge-to-Cloud architecture (NVIDIA Jetson NX, IXION, Snap7), the system monitors plant telemetry every 3 seconds and inlet levels every 3 minutes to maintain a real-world digital twin of the network.

A predictive module forecasts future user behavior via historical analysis, allowing a Deep Reinforcement Learning (DRL) agent to optimize collection policies. This agent balances energy efficiency with saturation avoidance by dynamically adjusting fan speeds and scheduling collections during low-cost energy windows.

Currently operational, the software has achieved a 3-4% reduction in energy consumption and improved service reliability. This deployment proves that integrating predictive AI with industrial hardware delivers tangible sustainability gains and superior urban waste management.