Apple is famously discrete when it comes to its research and future plans, especially in areas where they don’t have a dominant or powerful position. So it was a surprise to see their first research and plans concerning self-driving technology appear online. The paper, published on online journal arXiv, gives us a first glimpse into their work for the car industry #automagic
The paper talks about a mapping system that could be put to good use in the robotics, auto or mixed reality industry. It’s been a known fact that Apple’s original plan for designing a driverless car has been put aside only to focus on the self-driving technology other carmakers might need. In this endeavor, they appear to be focusing on 3D object detection with a software called VoxelNet.
This program is supposed to be more effective than others on the market because it uses only light-based LIDAR tech to detect 3D objects. Those sensors bounce lasers off objects in the immediate proximity, creating a 3D model of the environment in the time. The depth information collected is far more precise than the one offered by standard cameras but it’s not perfect. The maps produced are often sparse and need multiple systems to put them together right.
VoxelNet could compress all these processes and create a single neural network, increasing the overall efficiency of the system. Of course, this is only research at this point and the finalized Project Titan could look very differently. Nevertheless, it’s a hint regarding Apple’s view of the automotive industry and autonomous technology and the direction Tim Cook and his team might take in the next five years.