MIDAR: Mimicking LiDAR Detection for Traffic Applications with a Lightweight Plug-and-Play Model

2025-08-04

AI Summary

High-fidelity perception modeling and large-scale scalability remain mutually exclusive challenges in multi-vehicle collaborative perception simulation. Method: This paper proposes MIDAR—a lightweight, plug-and-play model that synthesizes realistic 3D object detection outputs approximating actual LiDAR detections, using only vehicle-level states (position, dimensions, heading) from microscopic traffic simulators (e.g., SUMO), without requiring raw point clouds. MIDAR explicitly constructs an RM-LoS graph to model multi-hop line-of-sight occlusion relationships and employs a GRU-enhanced APPNP graph neural network to propagate occlusion effects, jointly predicting true positives and missed detections. Contribution/Results: On nuScenes, MIDAR achieves an AUC of 0.909 when emulating CenterPoint detection performance. It demonstrates high fidelity in modeling individual vehicle perception states across two collaborative perception tasks, establishing—for the first time—an efficient and accurate bridge between microscopic traffic simulation and perception simulation.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Planning, Routing, and Scheduling: Model-Based Reasoning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

As autonomous driving (AD) technology advances, increasing research has focused on leveraging cooperative perception (CP) data collected from multiple AVs to enhance traffic applications. Due to the impracticality of large-scale real-world AV deployments, simulation has become the primary approach in most studies. While game-engine-based simulators like CARLA generate high-fidelity raw sensor data (e.g., LiDAR point clouds) which can be used to produce realistic detection outputs, they face scalability challenges in multi-AV scenarios. In contrast, microscopic traffic simulators such as SUMO scale efficiently but lack perception modeling capabilities. To bridge this gap, we propose MIDAR, a LiDAR detection mimicking model that approximates realistic LiDAR detections using vehicle-level features readily available from microscopic traffic simulators. Specifically, MIDAR predicts true positives (TPs) and false negatives (FNs) from ideal LiDAR detection results based on the spatial layouts and dimensions of surrounding vehicles. A Refined Multi-hop Line-of-Sight (RM-LoS) graph is constructed to encode the occlusion relationships among vehicles, upon which MIDAR employs a GRU-enhanced APPNP architecture to propagate features from the ego AV and occluding vehicles to the prediction target. MIDAR achieves an AUC of 0.909 in approximating the detection results generated by CenterPoint, a mainstream 3D LiDAR detection model, on the nuScenes AD dataset. Two CP-based traffic applications further validate the necessity of such realistic detection modeling, particularly for tasks requiring accurate individual vehicle observations (e.g., position, speed, lane index). As demonstrated in the applications, MIDAR can be seamlessly integrated into traffic simulators and trajectory datasets and will be open-sourced upon publication.

Problem

• Bridging gap between microscopic traffic simulators and LiDAR detection modeling
• Approximating realistic LiDAR detections using vehicle-level simulator features
• Enabling scalable perception modeling for multi-AV cooperative systems

Innovation

• Mimics LiDAR detection using vehicle-level features
• Uses GRU-enhanced APPNP for feature propagation
• Integrates seamlessly with traffic simulators