Identifying effective trajectory predictions under the guidance of trajectory anomaly detection model

Trajectory Prediction (TP) is an important research topic in computer vision and robotics fields. Recently, many stochastic TP models have been proposed to deal with this problem achieved better performance than the traditional deterministic trajectory outputs. However, these can generate a number of future trajectories different qualities. They are lack self-evaluation ability, that is, examine rationality their prediction results, thus failing guide users identify high-quality ones from candidate results. This hinders them playing best real applications. In paper, we make up for defect propose TPAD, novel evaluation method based on Anomaly Detection (AD) technique. firstly combine Automated Machine Learning (AutoML) technique experience AD field automatically design effective model. Then, utilize learned model predicted trajectories, screen out good results users. Extensive experimental demonstrate TPAD effectively near-optimal improving models’ practical application effect.