Proposal of an object detection method using point cloud data independent of color and brightness,with implementation evaluation on a hexapod robot

Abstract

The objective of this study is to develop a method that enables object detection independent of color or material, even with a limited dataset. This method is particularly effective for artificial structures such as stairs and steps, and it also demonstrates a certain level of effectiveness against new obstacles not included in the training. As a result, using a relatively small dataset based on four objects and a total of thirty-point clouds, we were able to demonstrate consistency and accuracy in object detection using a model that combines point cloud data from a depth camera with a CNN. This method suggests the possibility of overcoming the challenge of data scarcity and implementing practical object detection systems even in small organizations.