通过分段扩散搜索提升机器人长程操作成功率
FORGE-Tree: Diffusion-Forcing Tree Search for Long-Horizon Robot Manipulation
- 分阶段局部去噪，只优化当前关键段
- 搜索预算增加时成功率提升13.4~17.2个百分点
- 适合计算资源有限的长任务部署场景
长时程机器人操作对视觉-语言-动作（VLA）策略仍具挑战，因漂移和暴露偏差导致误差累积。现有方法常以固定超参数全轨迹去噪，缺乏在狭窄空间动态分配额外推理算力的能力。本文提出FORGE-Tree，一个可插拔控制层，结合阶段对齐的扩散强迫（DF）头与测试时蒙特卡洛树扩散（MCTD）。在冻结VLA编码器前提下，DF将时间步对齐至子任务阶段；推理时仅部分去噪目标段，其余标记保持冻结，将轨迹优化转化为一系列局部编辑。随后利用蒙特卡洛树扩散选择下一需优化段。场景图提供扩展先验与几何关系感知评分，实现树状结构去噪，性能随搜索预算提升而增长，同时保留已执行前缀。在LIBERO数据集上，FORGE-Tree相较原生VLA基线（使用OpenVLA与Octo-Base）成功率达13.4~17.2个百分点提升，且在相近算力预算下表现稳定，尤其在长时程变体中优势明显。
原文摘要 · Abstract (English)
Long-horizon robot manipulation tasks remain challenging for Vision-Language-Action (VLA) policies due to drift and exposure bias, often denoise the entire trajectory with fixed hyperparameters, causing small geometric errors to compound across stages and offering no mechanism to allocate extra test-time compute where clearances are tight. To address these challenges, we introduce FORGE-Tree, a plug-in control layer that couples a stage-aligned Diffusion Forcing (DF) head with test-time Monte Carlo Tree Diffusion (MCTD). With a frozen VLA encoder, DF aligns timesteps to subtask stages; during inference we partially denoise only a target segment while keeping other tokens frozen, turning trajectory refinement into a sequence of local edits. We then apply Monte Carlo Tree Diffusion to select the next segment to refine. A scene graph supplies priors for expansion and geometry relation-aware scoring for rollouts, yielding tree-structured denoising whose performance scales with search budget while preserving the executed prefix. Evaluation on LIBERO, FORGE-Tree improves success rate by 13.4 to 17.2 pp over the native VLA baselines with both OpenVLA and Octo-Base. Gains remain consistent under comparable compute budgets, especially on long-horizon variants. Videos available at: https://taco-group.github.io/FORGE-Tree/
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。