This repository offers source codes and ROS configuration files to run a robot capable of navigation, objects detection and recognition.
Ros package for object detection and recognition.
Package contains a node which can detect objects in front of a sensor. By using tensorflow recognition node from image_recognition package (tue-robotics) tries to recognize trained objects.
Before running detector_pkg, train neural network like in this tutorial image_recognition. Then start the tensorflow node:
rosrun tensorflow_ros object_recognition_node _graph_path:=<path/to/the/graph/from/training/output_graph.pb> _labels_path:=<path/to/the/labels/from/training/output_labels.txt>
Run detector node:
rosrun detector_pkg detector
ROS package implementing actionlib client to frontier_exploration server. Needed frontier_exploration launch file is available in the launch directory.
All necessary launch files to run navigation stack and whole object finder demo are available in the launch directory.