{"version":1,"type":"story","url":"https://digestai.news/story/cityplanner-a-new-sandbox-agent-for-urban-planning","json":"https://digestai.news/story/cityplanner-a-new-sandbox-agent-for-urban-planning.json","markdown":"https://digestai.news/story/cityplanner-a-new-sandbox-agent-for-urban-planning.md","slug":"cityplanner-a-new-sandbox-agent-for-urban-planning","headline":"CityPlanner: A New Sandbox Agent for Urban Planning","summary":"Urban planners face complex challenges in selecting feasible actions from large candidate spaces, often constrained by cost and service quality. Existing methods are effective but depend on task-specific representations and constraint handling. Researchers have introduced CityPlanner, a new sandbox-agent framework designed to address these issues. The framework includes UrbanSandbox, an environment where agents can inspect task files, generate plans, run evaluators, and revise decisions based on feedback. To make learning more tractable, the researchers propose atomic-task reinforcement learning, which breaks down long sandbox trajectories into BuildPlan (for initial construction) and ImprovePlan (for refinement). Experiments show that CityPlanner outperforms heuristic methods and general LLM agents. The framework is available at <https://anonymous.4open.science/r/co-agent-C1C8>.","keyPoints":["CityPlanner introduces UrbanSandbox for executable urban planning.","Atomic-task reinforcement learning decomposes long trajectories into BuildPlan and ImprovePlan.","Experiments demonstrate CityPlanner's superior performance over existing methods."],"whyItMatters":"This framework could significantly improve the efficiency and effectiveness of urban planning processes, potentially leading to better-designed cities and more sustainable urban development strategies.","category":{"slug":"agents","name":"Agents & Tools","url":"https://digestai.news/category/agents"},"entities":{"companies":[],"models":[],"people":[]},"firstPublishedAt":"2026-09-11T04:00:00Z","updatedAt":"2026-09-11T04:00:00Z","sourceCount":1,"hasPrimarySource":true,"sources":[{"outlet":"arXiv cs.AI","title":"CityPlanner: A Sandbox Agent for Executable Urban Planning","url":"https://arxiv.org/abs/2609.09578","publishedAt":"2026-09-11T04:00:00Z","type":"primary","primary":true,"lead":true}],"sourceNotes":null,"discussions":[],"thread":null,"cite":{"text":"Digest AI, \"CityPlanner: A New Sandbox Agent for Urban Planning\", 11 September 2026, https://digestai.news/story/cityplanner-a-new-sandbox-agent-for-urban-planning","publisher":"Digest AI","title":"CityPlanner: A New Sandbox Agent for Urban Planning","datePublished":"2026-09-11T04:00:00Z","url":"https://digestai.news/story/cityplanner-a-new-sandbox-agent-for-urban-planning"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}