Tim Cook has publicly commented on Apple’s work in autonomous systems before, and a new research paper from two Apple research scientists dives deeper into the company’s efforts. The paper explains how Apple is using a combination of LiDAR with other technologies for 3D object detection that represents the future.
Sylvania HomeKit Light Strip
The paper is authored by Yin Zhou, an AI researcher at Apple, and Oncel Tuzel, a machine learning research scientist at the company. Both have joined Apple within the last two years. Below are just some broad highlights, read the full paper here.
The paper explains how accurate detection of objects in 3D point clouds can be used in autonomous navigation, housekeeping robots, and more:
Accurate detection of objects in 3D point clouds is a central problem in many applications, such as autonomous navigation, housekeeping robots, and augmented/virtual reality. To interface a highly sparse LiDAR point cloud with a region proposal network (RPN), most existing efforts have focused on hand-crafted feature representations, for example, a bird’s eye view projection.
In this work, we remove the need of manual feature engineering for 3D point clouds and propose VoxelNet, a generic 3D detection network that unifies feature extraction and bounding box prediction into a single stage, end-to-end trainable deep network.
Furthermore, it shows how the aforementioned technology can be used in LiDAR-based car, pedestrian, and cyclist detection benchmarks. Specifically, the paper presents an alternative to hand-crafted feature representations in LiDAR-based 3D detection:
Most existing methods in LiDAR-based 3D detection rely on hand-crafted feature representations, for example, a bird’s eye view projection. In this paper, we remove the bottleneck of manual feature engineering and propose Vox- elNet, a novel end-to-end trainable deep architecture for point cloud based 3D detection.
Our approach can operate directly on sparse 3D points and capture 3D shape information effectively. We also present an efficient implementation of VoxelNet that benefits from point cloud sparsity and parallel processing on a voxel grid.
Zhou and Tuzel believe that their experiments represent the future of 3D object detection, providing better results than other technologies when detecting cars, cyclists, and pedestrians “by a large margin.”
Our experiments on the KITTI car detection task show that VoxelNet outper- forms state-of-the-art LiDAR based 3D detection methods by a large margin. On more challenging tasks, such as 3D detection of pedestrians and cyclists, VoxelNet also demonstrates encouraging results showing that it provides a better 3D representation
The full paper is definitely worth a read and offers a rare insight into Apple’s work on autonomous systems. Check it out here.