Autonomous Exploration of mobile robots

• This research focuses on autonomous exploration of mobile robots in unknown environments, using Karto SLAM for real-time mapping and A* search for efficient path planning.

+ It employs an improved DQN algorithm, incorporating Convolutional Neural Networks (CNNs) for environmental feature extraction, resulting in the D3QN PER algorithm for effective target point decision-making.

• The algorithm's effectiveness is verified in a ROS and Gazebo-based simulation environment, including modeling different environments and robots, training various decision models, and evaluating their exploration efficiency and adaptability in different test scenarios.

• Project Document: [PDF]