论文arXiv 2611.1026412 天前
LOWP规划器摒弃稠密BEV,参数量减47%且推理提速1.7倍
本文由 AI 根据公开内容整理,可能有误,请以原文为准:Look Only Where You Plan for Real-Time End-to-End Urban Driving(arXiv 2611.10264)
核心内容
针对强驾驶规划器中稠密鸟瞰图(BEV)构建占据约60%参数的瓶颈,研究提出LOWP(Look Only Where You Plan)规划器。该方法摒弃全场景BEV重建与物体/地图监督,仅关注规划路径相关区域。在nuScenes开环测试中,LOWP将平均L2误差从0.302m优化至0.296m,同时模型参数量仅为稠密方案的53%,在同GPU下推理速度提升1.7倍,显著降低了端到端规划的算力门槛。
技术上,LOWP设计两组各16个可学习查询,利用相机标定信息直接读取图像特征:一组基于投影3D锚点采样,另一组基于视线射线注意力机制,其中锚点读取器在转弯场景表现更优。在NAVSIM伪仿真中获85.9 PDMS评分;Bench2Drive闭环测试218条路线得分65.5。研究特别区分了实测延迟与硬件估算值,指出后者无法建立通用硬件排名基准,强调了真实系统部署中实测数据的重要性。
关键看点
- 稀疏特征读取:采用两组16个可学习查询替代稠密BEV,通过3D锚点采样与射线注意力直接提取图像特征,免除显式建图与物体检测监督。
- 效能量化对比:在nuScenes数据集上,以53%的参数量实现1.7倍推理加速,L2误差微降至0.296m,证明紧凑表征足以支撑高精度规划。
- 多场景评测验证:NAVSIM伪仿真PDMS达85.9;Bench2Drive闭环218条路线得分65.5,但存在刹车导致的静止故障,释放规则收益尚不明确。
- 延迟评估基准:明确区分实测推理延迟与硬件转移估算值,指出非统一硬件下的估算无法作为有效排名依据,强调具身系统需关注实测指标。
工程与产业观察
该工作为具身智能实时规划提供了重要的工程范式参考:通过任务驱动的稀疏特征采样替代全局稠密感知,可在保持性能前提下大幅降低端侧算力需求。对于机器人导航与移动操作算法研发,这种“按需感知”思路有助于解决高维视觉输入与有限机载算力的矛盾,但需注意闭环控制中的长尾失效案例及实测延迟基准的建立。
英文摘要 (Abstract)
Dense bird's-eye-view (BEV) construction accounts for 60% of the parameters in a strong recent driving planner. This paper presents LOWP, a planner that looks only where it plans: two sets of sixteen learnable queries read the cameras through the known calibration, one by sampling image features at projected 3D anchors and one by attending to features labeled with their viewing rays, so no dense BEV is built and no object or map is supervised. Evaluations on open-loop nuScenes, pseudo-simulated NAVSIM and closed-loop Bench2Drive show competitive planning with a compact representation. On nuScenes, average L2 changes from 0.302 to 0.296 m with 53% of the dense counterpart's parameters and 1.7 times its speed on the same GPU; the anchor-based reader performs better on turns than the global reader. On NAVSIM, scoring the planner's own trajectories alongside a vocabulary gives 85.9 PDMS without retraining. The closed-loop control configuration scores 65.5 on 218 routes, with a brake-induced standstill failure whose release rule has an inconclusive aggregate benefit. We distinguish measured latency from hardware-transferred estimates, which do not establish a common-hardware ranking.
作者:Haohua Que, Tianle Zhu, Qianyi Wu, Zhipeng Bao, Handong Yao
外部来源
本页事实对应的公开页面,以原文为准。(点击将离开本站)