IROS

Gating Before Commitment: Anticipating Intent Divergence to Prevent Post-Interaction Decision Failures in Autonomous Driving

承诺前的门控：预测意图分歧以防止自动驾驶中交互后的决策失败

Cong Xu, Ravi Sankar

机构 * University of South Florida（南佛罗里达大学） 纠错

AI总结 该研究提出承诺前的门控决策层，通过语言引导意图模块检测自动驾驶车辆交互中的意图分歧，可修复规划、降低误触发率，其对故障的最快检测和不确定性否决作用获实验支持。

Comments 8 pages, 4 figures. Submitted to the 16th Workshop on Planning, Perception and Navigation for Intelligent Vehicles (PPNIV) at IROS 2026. Supplementary video included as ancillary material

DESCENT: Directed Edge Scene Encoding for Airport Surface Movement Prediction

DESCENT：面向机场地面运动预测的有向边场景编码

Alexander Prutsch, David Schinagl, Horst Possegger

机构 * Graz University of Technology（格拉茨工业大学） ； Institute of Visual Computing（视觉计算研究所） 纠错

AI总结 该研究针对机场地面运动预测领域，提出基于Transformer的DESCENT架构，结合PRS上下文采样与检测Transformer解码器，在Amelia-10基准上较现有最优基线实现显著性能提升，尤其在安全关键场景表现突出。

Comments IROS 2026. Project page at this https URL (https://github.com/a-pru/descent)