AeroEval: Staged Program and Execution Validation for AI-Generated Drone Missions

Key points

• Large Language Models (LLMs) can generate drone programs from natural-language mission descriptions, but syntactically valid programs may still violate user intent, environmental constraints, and mission-level behavior.1
• AeroEval combines deterministic program analysis with context-grounded LLM agents.1
• In a stagewise ablation study, our Code and Trajectory Validators by themselves achieve mean run-level success rates of 44% and 56%, respectively, while the full AeroEval pipeline achieves 88%; the stages detect complementary failures in program structure, API usage, mission intent, obstacle avoidance, altitude, coverage, and event-driven transitions and the guided regeneration corrects for them.1
• These results demonstrate the benefit of combining program-level and execution-grounded agentic validation for AI-generated drone applications in the evaluated environment.1

Sources (1)

• [1]AeroEval: Staged Program and Execution Validation for AI-Generated Drone Missions
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 7, 09:49 AM

  We present AeroEval, an agent-assisted middleware for staged validation of AI-generated drone missions.

  Large Language Models (LLMs) can generate drone programs from natural-language mission descriptions, but syntactically valid programs may still violate user intent, environmental constraints, and mission-level behavior.

Extractive summary: sentences quoted from the sources.