KING Cross-Embodiment Kinematics Training and Adaptation Data
KING论文描述的混合运动学数据：Gazebo自动生成2WD/4WD滑移转向、3WD/4WD阿克曼、四足和六足六类形态语料，并随机化连杆、轮径、IMU位置与测量噪声；真实适配部分为Rover Mini和Unitree Go1各约1分钟平坦室内运动。项目页和论文可核验使用边界，但没有数据下载、总样本/小时、格式、划分、校准包或许可证。
- 开放状态
- 项目已核验·制品未开放
- 许可证
- No dataset license published
- 商业使用
- not established
- 人工评审结论
- paper_and_project_verified_mixed_sim_real_kinematics_data_not_released
- 机器人
- simulated 2WD skid-steer、simulated 4WD skid-steer、simulated 3WD Ackermann、simulated 4WD Ackermann、simulated quadruped、simulated hexapod、Rover Robotics Rover Mini、Unitree Go1
- 任务
- body-twist and contact-pose regression for cross-embodiment proprioceptive odometry
- 模态
- URDF/embodiment parameters、joint or wheel encoder position and velocity、IMU acceleration and angular velocity、contact state、body twist target、contact pose target、LiDAR-IMU SLAM reference for the short real adaptation traces
- 格式
- not released
- 采集/生成方式
- Gazebo for simulated collection
- 已知问题
- No downloadable files, manifest, total scale, timestamps, frame rate, format, calibration package, privacy statement, license, code or weights. Real data is only about one minute per platform on a flat indoor floor, and reference twist is derived from Livox MID-360 LiDAR-IMU SLAM.