A recent study explores integrating machine learning technologies into Autonomous Mobile Robots (AMRs) to enhance efficiency in dynamic environments.

The research focuses on developing an AMR incorporating SLAM, odometry, and deep learning algorithms for artificial vision, utilizing high-performance Jetson Nano embedded systems.

Obstacle avoidance and path planning are achieved using the AMCL algorithm.

Two CNNs, ResNet18 and YOLOv3, are employed for real-time object detection and scene perception, with ResNet18 achieving high accuracy and YOLOv3 demonstrating strong performance in object detection.

The study highlights the importance of integrating artificial intelligence into robotic systems for improved navigation and adaptability in complex environments.

Reinforcement Learning (RL) is emphasized for its effectiveness in path planning and navigation in AMRs.

The research also explores the use of deep learning algorithms and RL for enhanced navigation in uncertain environments, proposing integration with SLAM, odometry, and artificial vision for path planning using ROS and Jetson Nano.

The significance of RL algorithms in improving path planning and navigation for AMRs is underscored.

The study showcases trends in the field and emphasizes the importance of hardware components like GPUs, embedded systems, and sensors for robot functionality and performance.

Modifications to the power supply system are discussed to enhance autonomy and safety in AMRs, ensuring reliable communication between modules.