Expert Demonstrations for Autonomous Navigation of a Ground Vehicle

该数据集包含了1万个使用基于碰撞锥的控制障碍函数生成的专家演示，用于自主导航任务。这些演示是通过采样初始机器人状态、障碍物参数（位置、速度和半径）以及目标位置来收集的。该数据集旨在评估在变化的障碍物条件下安全性和可达性的表现，重点关注泛化能力和性能。规模上，数据集包含了1万个专家演示，任务则是地面车辆的自主导航。

This dataset comprises 10,000 expert demonstrations generated via collision cone-based control barrier functions for autonomous navigation tasks. These demonstrations are collected by sampling initial robot states, obstacle parameters including position, velocity and radius, as well as goal positions. This dataset aims to evaluate the performance of safety and reachability under varying obstacle conditions, with a focus on generalization capability and performance. In terms of scale, the dataset includes 10,000 expert demonstrations, and the target task is autonomous navigation for ground vehicles.