Bruxelles-Propreté | A positive assessment of AI for a cleaner region

Bruxelles-Propreté has completed the first phase of testing an AI-powered system developed by a Swiss startup to assess street cleanliness across 337 km of regional roads. Four waste collection trucks were equipped with cameras and onboard computers to identify and classify common types of litter—like cigarette butts, packaging, and leaf piles—and map their distribution using a Clean City Index (CCI). Initial results show CCI scores ranging from 3.3 to 4.2 (it indicates that most streets were moderately to very clean), and the data proved valuable in reallocating cleaning resources from cleaner to dirtier areas more efficiently. 

Interestingly, tests showed that reducing cleaning frequency by up to 60% had little impact on overall cleanliness, freeing up staff for roughly 30 additional cleaning rounds. While further testing is needed, the technology offers a promising, results-driven tool to optimize cleaning operations without replacing on-the-ground observation. Encouraged by the outcomes, the agency is preparing a second evaluation phase involving 2–4 municipalities to measure the effects of actions like communication campaigns, enforcement, and changes to waste bin placement. 

Source: www.arp-gan.be (in French)