An Uncertainty Aware Interaction and Intent aware Transformer for Multimodal Trajectory Prediction of Dynamic Road Objects in Autonomous Vehicles
Contributors
Dr. PRAVEEN CHANDAR J
Dr.Shish Ahmad
Keywords
Proceeding
Track
General Track
License
Copyright (c) 2026 Sustainable Global Societies Initiative

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
Abstract
Autonomous vehicle is one of the emerging Edge AI Technology nowadays and it is being utilized as public transport in developed countries. It is very essential to assure the decision making capability and operational safety of the vehicle during uncertainty conditions. A novel moving object prediction approach is proposed in this research work based on multimodal trajectory prediction. Object identification is classified in to stable object identification and moving object detection which contribute in decision making. In this proposed approach interaction and intent aware model such as an uncertainty aware transformer- based trajectory prediction approach is proposed to identify the dynamic moving objects. The sensor module gives the data to historical trajectory module. The data is categorized as motion features, scene context and intent features. After Interaction encoding, the data is given to spatio-temporal transformer. The multimodal encoder splits the data in to various modes and estimates the probability, uncertainty and risk attainment level. Based on this risk aware trajectory assessment, decision making and vehicle planning is executed. The proposed approach is experimented on Google Colab platform and results are observed. The results demonstrate that the proposed approach produces the highly accurate prediction results compared with other approaches.