Identifying Unknown Instances for Autonomous Driving
This addresses the critical need for robots to handle diverse real-world objects beyond predefined categories, though it is incremental as it builds on existing deep learning methods for perception.
The paper tackles the problem of identifying unknown object instances in autonomous driving by developing an open-set instance segmentation algorithm for point clouds, achieving effective segmentation of both known and unknown classes as validated on two large-scale self-driving datasets.
In the past few years, we have seen great progress in perception algorithms, particular through the use of deep learning. However, most existing approaches focus on a few categories of interest, which represent only a small fraction of the potential categories that robots need to handle in the real-world. Thus, identifying objects from unknown classes remains a challenging yet crucial task. In this paper, we develop a novel open-set instance segmentation algorithm for point clouds which can segment objects from both known and unknown classes in a holistic way. Our method uses a deep convolutional neural network to project points into a category-agnostic embedding space in which they can be clustered into instances irrespective of their semantics. Experiments on two large-scale self-driving datasets validate the effectiveness of our proposed method.