ntnu-arl/deepvl-training-data

DeepVL训练数据集包含用于水下机器人速度学习的训练和测试数据集。该数据集是通过在泳池和特隆赫姆峡湾手动驾驶水下机器人收集的。数据集分为训练数据（包含traj1到traj12的轨迹）和测试数据（包含traj1到traj2的轨迹）。每个轨迹文件夹中包含了不同类型的数据文件，如Alphasense IMU数据、估计的IMU偏差、飞行控制单元IMU数据、重力向量、电机命令PWM信号、方向矩阵、地面真实里程计数据等，所有文件均为.npy格式，可以用numpy加载和解析。

The DeepVL training dataset consists of training and testing datasets used for underwater robot velocity learning. The dataset was collected by manually piloting an underwater robot in a pool and in the Trondhiem fjord. It includes training data (containing trajectories from traj1 to traj12) and testing data (containing traj1 to traj2). Each trajectory folder contains various types of data files such as Alphasense IMU data, estimated IMU biases, flight control unit IMU data, gravity vector, motor command PWM signals, orientation matrices, ground-truth odometry, etc., all in `.npy` format and can be loaded and parsed using numpy.