We present OptObstacles, a test case generator for unmanned aerial vehicles (UAVs) that leverages metaheuristic search over obstacle configurations (sizes, positions, and rotation angles). Our approach searches for challenging test cases by optimizing for objective functions designed to represent (1) constraints on geometric validity (e.g., non-overlapping obstacles), (2) simulation outcomes (e.g., minimum distance between UAV and obstacles), and (3) custom heuristics (e.g., potentially favorable positions relative to the expected UAV trajectory). In our preliminary testing, our approach generates challenging scenarios, which yield a higher rate of soft failures (compared to full failures). OptObstacles is submitted for participation in the UAV testing competition at SBFT 2025, and our code is publicly available at https://github.com/zhekai-jiang/UAV-Testing-Competition.