We’re safe for now; robots are still very complicated, especially keeping them running smoothly. An excellent published PhD thesis highlights this complexity and presents a detailed design for real-time monitoring and control of various technical aspects of industrial robots. This book crosscuts many domains touched by industrial robots, including signal processing to detect faults, software for embedded monitoring, and wireless data collection and fault processing. Although it is written about robots, many of the described techniques are also applicable to other applications. Condition monitoring has traditionally been an offline process, but new approaches help turn those into online monitoring and robot control. This supports various root cause diagnostics, as well as predictive and preventative maintenance solutions.
This book is part of the “Springer Theses” series, and represents the author’s doctoral dissertation. As such, the format includes both important relationships between the author’s work and others, and a robust literature review.
Numerous sections discuss condition monitoring concepts including model-based, data-driven, and model-free approaches. Industrial robots deal with many moving parts and are continually running in a changing environment. So to monitor robot-specific faults, vibration monitoring is a critical approach. This information is fed to two stages of the author’s intelligent condition monitoring algorithms: (1) fault detection and (2) fault diagnostics. With appropriate root-cause analysis design, these algorithms help keep the robots running. The author’s design then passes that information to the analysis phase, shown in the book through wired or wireless data transmission and processing.
The thesis part of the book starts by discussing all the signal processing approaches needed to actually monitor a complicated industrial robot. A rationale for this is that embedding sensors everywhere isn’t practical (such as the tooth wear on gears). Instead, the research shows how to exploit various signal analysis techniques. This includes time-domain, frequency-domain, and time-frequency signal analysis techniques.
To showcase the author’s new intelligent condition monitoring, an actual PUMA 560 Industrial Robot is used. The book shows how some of these condition monitoring approaches would fit with this type of robot. It looks at things like the transmission faults and data acquisition approaches supported by the robot hardware. Here is where the author describes in great detail how a lot of moving parts make robots powerful but also highlights where faults are prevalent. The mechanical moving elbow joint, the gears, and other areas are discussed in detail. These provide a lot of context both on how faults are detected and on how they can be simulated to be ready to offer solutions for suspected faults.
While a lot of the robot information is needed for the author’s application, the actual condition monitoring design is applicable to many other application domains; fault detection and fault diagnosis stages are common patterns. Much of this involves charting the real-time diagnosed information against both norms, and potential simulated errors, maybe gathered from a training event.
For software designers, the last two chapters are exciting. These focus on the design of the embedded system used for robot health monitoring and the software to affect the collaboration between the various sensor nodes. The author shows how a wireless approach will provide for a flexible and dynamic approach to data acquisition. Here, an embedded controller manages fault acquisition and broadcasts that information out to the various analysis processors.
The book provides great information for those interested in exploring similar solutions for online monitoring of complicated systems, from robots to airplane part fault diagnostics. In particular, various wireless sensor network approaches and topologies are discussed with respect to scalability and maintainability characteristics. The sensor node is the end device that collects the information from the robot, and as the author describes, combines with a microcontroller using wireless communication approaches. Standards from Wi-Fi to Bluetooth to ZigBee are available, all with different processing speeds, bandwidth constraints, and power consumption. The data from the sensor nodes are then distributed via wireless technology to the base station for processing. Various elements of the embedded software design are discussed, including the system testing and validation steps. Finally, power consumption issues are analyzed to help designers understand the sensor node constraints when dealing with long running systems.
In all, this excellent published thesis is very methodically organized and researched. The information is pertinent to a wide level of readers in fields such as hardware design, embedded computing, and distributed and wireless communication. Then, to keep a complicated system running, the main aspect of the work describes how all these can be combined to detect, collect, analyze, and hopefully prevent faults before they occur, keeping the next generation of robots continually running smoothly.
