On Exploring Input Resolution Scaling For Anytime LiDAR Object Detection

2026-07-09

AI Summary

This work addresses the challenge of balancing latency and accuracy in LiDAR-based 3D object detection under dynamic runtime conditions to meet real-time requirements. The authors propose a single-model, multi-resolution inference framework that enables anytime computation through dynamic input resolution scaling and incorporates a deadline-aware scheduler to select the optimal resolution at runtime. Key innovations include deploying only one DNN model to eliminate redundancy, designing a high-precision runtime execution time predictor tailored for irregular point clouds, and integrating pillar/voxel-based architectures for memory-efficient inference. Experiments on nuScenes demonstrate significant improvements over existing anytime methods, and evaluations in a simulated autonomous driving system show collision-free navigation with notably reduced unnecessary halts in complex environments.

Abstract

Making tradeoffs between execution latency and result utility (i.e., anytime computing) for adapting to dynamic operational requirements has been shown to enhance the performance of cyber-physical systems. In this work, we focus on enabling anytime computing for deep neural networks (DNNs) that process LiDAR point clouds for 3D object detection. We propose a novel method that enables multi-resolution inference for models that process point clouds as pillars or voxels, allowing the input to be dynamically scaled and processed at the resolution needed to meet timing requirements. Importantly, our memory-efficient approach requires the deployment of only a single DNN model, avoiding the need to deploy multiple models, each trained for a different input resolution. We also introduce a deadline-aware scheduler that selects the highest possible resolution for any given input by accurately predicting the execution time for all possible resolutions at runtime, which is challenging due to the irregularity of LiDAR point clouds. Experimental results on the nuScenes autonomous driving dataset demonstrate that our method significantly outperforms existing anytime computing approaches for LiDAR object detection. Finally, we deploy our approach in a simulated autonomous driving system, where it consistently enables collision-free navigation while avoiding unnecessary stalls caused by environmental complexity.

Problem

• anytime computing
• LiDAR object detection
• input resolution scaling
• 3D object detection
• cyber-physical systems

Innovation

• anytime computing
• LiDAR object detection
• multi-resolution inference
• deadline-aware scheduling
• point cloud processing