The concept of the smart factory involves autonomous mobile robots (AMRs) deployed to execute repetitive tasks, such as finding, sorting, and moving packages in large warehouse facilities. These robots reduce the need for employees to walk throughout the warehouse facility and increase the number of orders picked in an hour.
Prior to development of AMRs, automated ground vehicles (AGVs) were used in restricted and limited spaces using guidance methods such as white lines and magnetic tape. Advances in perception, motion planning, and control have enabled AMRs to sense the environment, detect obstacles, and make decisions to move from one place to another. These developments have significantly broadened the range of AMR solutions for many applications.
This ebook discusses AMR technologies and development challenges, and how you can use MATLAB® and Simulink® to design, simulate, and deploy AMR algorithms.
Examples of Autonomous Mobile Robot Applications
AMRs are used in industries such as automotive, industrial automation, and consumer electronics. The technology used to develop AMRs is also applicable to many other unmanned ground vehicles (UGVs), as well as self-driving vehicles. AMR and AGV applications include:
Autonomous operations in factories and production lines
Space exploration using rovers
Nursing care using service robots
Security patrolling using service robots
How Autonomous Mobile Robot Systems Work
Autonomous mobile robots (AMRs) operate using a combination of complex algorithms for functions including object detection, situational awareness, decision-making, and vehicle control. These algorithms are tightly coupled in a closed-loop feedback system that constrains system behavior.
Autonomous mobile robot (AMR) system.
Conventional Mobile Robots
Prior to the introduction of autonomous capabilities, mobile robots required visual monitoring because of their limited capacity for self-guided operations. Still in use for certain applications, these conventional mobile robots generally have the following characteristics:
Single-loop synchronous processing
Sensor responses based on one-dimensional time series signals
A limited number of actuators, sometimes only one
Architecture of conventional mobile robots.
Challenges with Autonomous Mobile Robots
AMR technologies must be able to manage the complexity of cognitive judgements that humans execute in the conventional systems. For example, the advanced obstacle detection and avoidance capability necessary for an AMR requires a combination of multidisciplinary technologies such as deep learning for object recognition, lidar point cloud processing, obstacle tracking by sensor fusion, map generation by simultaneous localization and mapping (SLAM), path planning, and path tracking control.
Because of these complexities, AMR development faces the following technical challenges:
Technology integration across multiple domains such as image processing, motion planning, and control
Multiple sensor and actuator combinations requiring coordinate transformations
High-resolution and multidimensional sensor processing (distributed and parallel)
Time synchronization and sensor fusion, including mixed synchronous and asynchronous processing to accommodate different sensor and actuator rates
Architecture of AMRs.
Autonomous Mobile Robot Development Using MATLAB and Simulink
MATLAB and Simulink are development and verification platforms that can be used to design, simulate, and deploy AMR systems. The benefits of using MATLAB and Simulink include:
Multidomain technology can be handled on a single platform
Complex sensor and actuator processing can be flexibly described
High-resolution, multidimensional sensor processing can be parallelized by multicore CPU or GPU
In the following chapters, you will learn how you can use MATLAB, Simulink, and related toolboxes to develop AMRs.