Multimodal Model Predictive Runtime Verification for Safety of Autonomous Cyber-Physical Systems

Alexis Aurandt, Phillip H. Jones, Kristin Yvonne Rozier, and Tichakorn Wongpiromsarn

This webpage contains further details and artifacts for reproducibility of the experiments in "Multimodal Model Predictive Runtime Verification for Safety of Autonomous Cyber-Physical Systems"

---

Open-Source GitHub Repos

LaneGCN: https://github.com/uber-research/LaneGCN

LAformer: https://github.com/mengmengliu1998/LAformer

mmTransformer: https://github.com/decisionforce/mmTransformer

Lane Transformer: https://github.com/mmdzb/Lane-Transformer

HiVT-64 and HiVT-128: https://github.com/ZikangZhou/HiVT

---

Argoverse 1 Motion Forecasting Dataset

Available here: https://www.argoverse.org/av1.html#forecasting-link

Note, the three seconds of prediction for the testing split is not publicly available as the testing dataset is utilized for the Argoverse leaderboard/competitions. Accuracy metrics self-reported or reported here: https://eval.ai/web/challenges/challenge-page/454/leaderboard/1279

---

mmTransformer Implementation

This repo is forked from the open-source repo and includes additional functionality for formatting and visualizing the data utilized within MMPRV: https://github.com/aaurandt/MMPRV_mmTransformer

---