4.1 A. KITTI Dataset | KITTI 数据集

4.1.1 paragraph

The KITTI dataset [2] contains stereo sequences recorded from a car in urban and highway environments.

KITTI数据集包含双目数据，这些数据从一个正在高速公路上行驶的车上采集到的。

The stereo sensor has a 54cm baseline and works at 10Hz with a resolution before rectification of $1392 \times 512$ pixels.

这个双目传感器有个小于54厘米的基线并且在在 $1392 \times 512$ 像素上，以10Hz的采样速率进行采样，

Sequences 00, 02, 05, 06, 07 and 09 contain loops.

其中序列00, 02, 05, 06, 07 和 09包含回环。

Our ORB-SLAM2 detects all loops and is able to reuse its map afterwards, except for sequence 09 where the loop happens in very few frames at the end of the sequence.

我们的ORB-SLAM2能够检测出回环并且能够地图重用，除了09序列以外，09序列的回环只发生在尾端少数的几帧当中。

image copy right belongs to Raúl Mur-Artal and Juan D. Tardós. ORB-SLAM2: an Open-Source SLAM System for Monocular, Stereo and RGB-D Cameras. ArXiv preprint arXiv:1610.06475

图像摘录自ORB-SLAM2: an Open-Source SLAM System for Monocular, Stereo and RGB-D Cameras. ArXiv preprint arXiv:1610.06475

表1两种SLAM在测试KITT，I数据的精度对比

Table I shows results in the 11 training sequences, which have public ground-truth, compared to the state-of-the-art Stereo LSD-SLAM [11], to our knowledge the only stereo SLAM showing detailed results for all sequences.

表1显示的这个结果11个训练数据，这是一个公开的真实数据，对比于原先的LSD-SLAM算法，我们展示了的双目SLAM系统测试数据结果。

We use two different metrics, the absolute translation RMSE $t_{abs}$ proposed in [3], and the average relative translation $t_{rel}$ and rotation $r_{rel}$ errors proposed in [2].

我们使用两个不同的米制，均方根误差 $t_{abs}$ 在论文[3]中提到，并且取平均相关 $t_{rel}$ 平移和 $r_{rel}$ 旋转误差在论文[2]，

Our system outperforms Stereo LSD-SLAM in most sequences, and achieves in general a relative error lower than 1%.

我们的系统在大多数序列当中比双目的LSD-SLAM要优秀很多，并且能够获得的相关误差低于1%。

The sequence 01, see Fig. 3, is the only highway sequence in the training set and the translation error is slightly worse.

这个序列01如图3所示，是一个高速公路的序列，作为训练集，以及转换误差。

Translation is harder to estimate in this sequence because very few close points can be tracked, due to highspeed and low frame-rate.

转换误差是在这个序列当中难以评估的，因为只有几个很近的点能够被侦测，由于很高的速度和较低的帧率。

However orientation can be accurately estimated, achieving an error of 0.21 degrees per 100 meters, as there are many far point that can be long tracked.

然而这个方向能够被精确的评估，获得的误差是每100米做0.21度。很多较远的点能够被检测，

Fig. 4 shows some examples of estimated trajectories.

如图4所示，显示了一些评估的例子。

image copy right belongs to Raúl Mur-Artal and Juan D. Tardós. ORB-SLAM2: an Open-Source SLAM System for Monocular, Stereo and RGB-D Cameras. ArXiv preprint arXiv:1610.06475

图像摘录自ORB-SLAM2: an Open-Source SLAM System for Monocular, Stereo and RGB-D Cameras. ArXiv preprint arXiv:1610.06475

Fig. 4. Estimated trajectory (black) and ground-truth (red) in KITTI 01, 05, 07 and 08.

图4 在KITTE数据集01,05,07和08数据集，估计轨迹（黑色线）和以及实际运动（红色线）。