YOU SAID:

However, this proclivity for sharp turns off of the observed trajectory isn't reflected in the final metrics, as both the Uniform RRT* baseline and the KDE RRT* baseline performed the best as compared to the other baselines and the proposed LSTM-MDN RRT* model. This is due to the evaluation metrics not including a metric for the overall ``smoothness'' or variance of the outputted trajectory. The evaluation metrics only rely on the overall accuracy of the outputted trajectory as compared to the actual trajectory, and evaluate based on the average error between the two trajectories, or the displacement of the final trajectory point to the estimated goal. The metrics for error calculation can be found in Section 4.5. That being said, the trajectory that has a higher overall accuracy while still being highly noisy and not smooth may end up being more helpful in the path planning module for an autonomous vehicle, as the overall goal is just to predict their position instead of having a more intricate prediction that better models the overall behavior.

INTO JAPANESE

ただし、観測された軌跡が急激に変化するこの傾向は、最終的な指標には反映されません。これは、Uniform RRT* ベースラインと KDE RRT* ベースラインの両方が、他のベースラインおよび提案された LSTM-MDN RRT と比較して最高のパフォーマンスを発揮したためです。 * モデル。これは、全体的な「スムーズさ」のメトリックが評価メトリックに含まれていないためです。

BACK INTO ENGLISH

However, this sharp change in the observed trajectory is not reflected in the final metric. This is because both the Uniform RRT* baseline and the KDE RRT* baseline performed best compared to other baselines and the proposed LSTM-MDN RRT.

INTO JAPANESE

ただし、観測された軌跡のこの急激な変化は、最終的なメトリックには反映されません。これは、Uniform RRT* ベースラインと KDE RRT* ベースラインの両方が、他のベースラインおよび提案された LSTM-MDN RRT と比較して最高のパフォーマンスを発揮したためです。

BACK INTO ENGLISH

However, this abrupt change in observed trajectory is not reflected in the final metric. This is because both the Uniform RRT* baseline and the KDE RRT* baseline performed best compared to other baselines and the proposed LSTM-MDN RRT.

INTO JAPANESE

ただし、観測された軌跡のこの急激な変化は、最終的なメトリックには反映されません。これは、Uniform RRT* ベースラインと KDE RRT* ベースラインの両方が、他のベースラインおよび提案された LSTM-MDN RRT と比較して最高のパフォーマンスを発揮したためです。

BACK INTO ENGLISH

However, this abrupt change in observed trajectory is not reflected in the final metric. This is because both the Uniform RRT* baseline and the KDE RRT* baseline performed best compared to other baselines and the proposed LSTM-MDN RRT.