Autonomous Garbage-Collecting Robot For Beaches With Deep Learning Approach and Improved Cleaning Technique

This research is to develop an autonomous garbage collecting robot using a deep learning approach and an improved cleaning technique. In the proposed methodology, the system can distinguish between garbage and non-garbage items using CNN technology, which is one of the deep learning approaches. By using a Pi camera, the system can obtain raw images of the surroundings and undergo the process of image recognition and feature extraction with the help of TensorFlow to perform object detection. Based on the bounding box surrounding the object, the system obtains the box’s coordinates and performs object tracking by detecting the object’s position and controlling the motors for position adjustment.

The cleaning technique implemented in the system uses a series of rotary blades for sweeping the garbage toward a mesh-net platform as the collecting mechanism. If the storage compartment is full, the movable gate will be opened for disposing of the garbage inside. By these, the robot can imitate human behavior of detecting, distinguishing, and collecting garbage.