Next-Best-View Estimation based on Deep Reinforcement Learning for Active Object Classification
This work addresses the challenge of insufficient information from single viewpoints in robotics, offering an incremental improvement for active object classification tasks.
The paper tackles the next-best-view problem for active object classification by using a robot arm to find optimal viewpoints that improve object identification, achieving more accurate predictions to better distinguish objects.
The presentation and analysis of image data from a single viewpoint are often not sufficient to solve a task. Several viewpoints are necessary to obtain more information. The next-best-view problem attempts to find the optimal viewpoint with the greatest information gain for the underlying task. In this work, a robot arm holds an object in its end-effector and searches for a sequence of next-best-view to explicitly identify the object. We use Soft Actor-Critic (SAC), a method of deep reinforcement learning, to learn these next-best-views for a specific set of objects. The evaluation shows that an agent can learn to determine an object pose to which the robot arm should move an object. This leads to a viewpoint that provides a more accurate prediction to distinguish such an object from other objects better. We make the code publicly available for the scientific community and for reproducibility.