为机器人导航设计新基准，关注定位精度与可重复性。
Task-driven SLAM Benchmarking For Robot Navigation
- 以任务为导向，用精度衡量SLAM性能，兼顾建图能力。
- 实测显示被动双目在室内精度接近激光雷达SLAM。
- 适合评估导航场景下SLAM表现，支持自定义环境测试。
移动助手机器人执行基于任务的导航时，需要依赖SLAM进行定位。现有SLAM基准测试忽视了可重复性（精度）的重要性，而这对实际部署至关重要。为此，本文提出任务驱动的SLAM基准方法——TaskSLAM-Bench，采用精度作为核心指标，考量SLAM的建图能力，并具备易于实现的条件。通过模拟与真实场景测试，揭示了现代视觉与激光雷达SLAM方案在导航性能上的特性。结果表明，在典型室内环境中，被动双目SLAM的定位精度可达到与激光雷达SLAM相当水平。TaskSLAM-Bench弥补了现有基准的不足，为导航导向的SLAM性能评估提供更丰富视角。公开代码支持配备适当传感器的机器人在自定义环境中进行现场测试。
原文摘要 · Abstract (English)
A critical use case of SLAM for mobile assistive robots is to support localization during a navigation-based task. Current SLAM benchmarks overlook the significance of repeatability (precision), despite its importance in real-world deployments. To address this gap, we propose a task-driven approach to SLAM benchmarking, TaskSLAM-Bench. It employs precision as a key metric, accounts for SLAM's mapping capabilities, and has easy-to-meet implementation requirements. Simulated and real-world testing scenarios of SLAM methods provide insights into the navigation performance properties of modern visual and LiDAR SLAM solutions. The outcomes show that passive stereo SLAM operates at a level of precision comparable to LiDAR SLAM in typical indoor environments. TaskSLAM-Bench complements existing benchmarks and offers richer assessment of SLAM performance in navigation-focused scenarios. Publicly available code permits in-situ SLAM testing in custom environments with properly equipped robots.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。