Achieving Real-Time LiDAR 3D Object Detection on a Mobile Device
This work is significant for the autonomous driving industry, as it enables real-time 3D object detection on edge-computing devices, which is crucial for practical deployment.
This paper addresses the challenge of real-time LiDAR 3D object detection on resource-constrained mobile devices for autonomous driving. The authors propose a compiler-aware unified framework that uses a generator Recurrent Neural Network (RNN) for automatic network enhancement and pruning search, enabling real-time inference on devices like the Samsung Galaxy S20 phone with competitive detection performance.
3D object detection is an important task, especially in the autonomous driving application domain. However, it is challenging to support the real-time performance with the limited computation and memory resources on edge-computing devices in self-driving cars. To achieve this, we propose a compiler-aware unified framework incorporating network enhancement and pruning search with the reinforcement learning techniques, to enable real-time inference of 3D object detection on the resource-limited edge-computing devices. Specifically, a generator Recurrent Neural Network (RNN) is employed to provide the unified scheme for both network enhancement and pruning search automatically, without human expertise and assistance. And the evaluated performance of the unified schemes can be fed back to train the generator RNN. The experimental results demonstrate that the proposed framework firstly achieves real-time 3D object detection on mobile devices (Samsung Galaxy S20 phone) with competitive detection performance.