用向量加速技术让多机器人规划毫秒级完成
Multi-Robot Motions in Milliseconds: Vector-Accelerated Primitives for Sampling-Based Planning

- 设计向量并行的多机器人运动验证与冲突检测原语
- 相比传统方法,规划速度最高提升1492倍,四机械臂任务均在1秒内完成
- 适合需要实时多机协同的工业机器人、移动平台等场景
本文将近期提出的向量加速运动规划(VAMP)框架扩展至多机器人场景,开发了两种向量加速原语:多机器人运动验证(MotVal)和首个冲突检测(FFC),充分利用多机器人环境中的SIMD并行性。纯多机器人运动验证测试中,验证时间提速超过1415倍。我们进一步改造五种代表性算法,使其采用这些新原语,构建出向量加速多机器人运动规划(VA-MRMP)系统。在机械臂、二维移动机器人及异构机器人团队任务中评估,相较FCL基准,规划速度最高提升1492倍。使用VA-MRMP后,五个规划器在四台Panda机械臂的问题上均实现亚秒级中位运行时间。
原文摘要 · Abstract (English)
In this paper, we extend the recent Vector-Accelerated Motion Planning (VAMP) framework to multi-robot motion planning. We develop two vector-accelerated primitives, multi-robot MotionValidation (MotVal) and FindFirstConflict (FFC), which exploit SIMD parallelism within the multi-robot domain. On pure multi-robot motion validation tests, this achieves over 1415X speedup in validation time. Additionally, we modify a representative set of algorithms to use these new primitives. We evaluate five vector-accelerated multi-robot motion planning (VA-MRMP) algorithms on manipulator, 2D mobile robot, and heterogeneous teams, observing planning speedups over FCL of up to 1492X. With VA-MRMP, all five planners attain subsecond median runtimes on problems with four Panda manipulators.
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。