Autonomous Mobile Robot Motor Control

(Source: SweetBunFactory/Stock.Adobe.com)

Autonomous Mobile Robots (AMRs) continue to be deployed in more warehousing and industrial applications at an increasing rate. Though artificial intelligence/machine learning (AI/ML) and vision/sensing systems are the intellectual core of an AMR, the muscle of an AMR is the motor control and electrical motor systems. Electrical motors, drivers, and control circuits are essential for an AMR to be mobile or perform tasks by physically manipulating targets. How well a motor control system and motor driver circuitry are implemented dictates the speed, reliability, efficiency, and precision of an AMR's mobility and physical manipulation capability.

This article will educate readers on the significance of motor control systems in AMR and further discuss the engineering considerations associated with designing motor control systems for AMR.

Overview of AMR Motion/Motor Control

Though there has been some change in standardized nomenclature, AMRs with or without manipulators, in most cases, are industrial mobile robots (IMRs). However, these nomenclature changes haven't changed the underlying mechanics of AMRs, simply the relative safety standards. As long as an AMR is not a low lift or high reach forklift primarily used for moving pallets using forks or tines, AMRs are currently governed by safety standards ANSI/RIA 15.08/15.0X in the US. 

These new standards recognize AMRs as collaborative robotic systems that work alongside and in the same vicinity as human workers. This new paradigm places a greater emphasis on safety and performance considerations on AMRs, which come as additional considerations to the already looming challenge of designing and developing an industrial robotic system. A key part of this is the motor control system since, without a reliable and responsive motor control system designed with functional safety features, an AMR is unlikely to be truly safe.

Each portion of a motor control system must meet performance expectations while being functionally safe, requiring the motor control system, drivers, and motors to operate with extreme levels of reliability and consistency. Other factors play a role in AMR motor control function, such as the sensing apparatus of the feedback loop and AI/ML algorithms, but those aspects of an AMR will not be highlighted here.

Motion/Motor Control Systems

There are typically multiple motor control systems in an AMR. Generally, at least one controls motion while others control each actuator/manipulator system mounted on the AMR. The purpose of the motor control systems is to take the AI/ML processed input from relevant sensors on the AMR and send signals to the motor driver circuits that carry the correct data to drive the motion and actuator/manipulator motors in the desired fashion. The motor control systems handle much of the functional safety aspects of an AMR's motion and actuation/manipulation functions. However, the motor drivers and electrical motors often need their own failsafe operational dynamics.

There are differences between a motion control system and a motor control system. The motion control system uses a variety of sensor input and higher-level control signals, such as trajectory planning, velocity planning, acceleration planning, interpolation algorithms, and kinematic conversions, to physically move an AMR's location. On the other hand, a motor control system is a fundamental system that controls a motor's response concerning input from the motor and motor control sensing apparatus. When it comes to an AMR, a motor control system could be a motion control system or a manipulation/actuation control system depending on the use case.

Generally, motion/motor control systems are either implemented on microcontrollers/digital signal processors (MCUs/DSPs) in software or on field-programmable gate arrays (FPGAs). MCUs/DSPs are a widespread solution for many motor control applications and robotics, and a wide range of options are available that include MCUs/DSPs with functional safety features. Many MCUs/DSPs for advanced motion/motor control applications also come with many other features, such as communication, signal processing, and motor driver output circuits built into the devices. These features ease development and represent a functional limit to an MCU/DSP. Suppose additional motor drivers, signal processors, communication systems, or other hardware circuits are needed. In that case, additional hardware must be developed to fill this gap, or a more complex/expensive MCU/DSP can be used instead that includes these features.

There are also various FPGA options for motor control, many of which enable the design of functionally safe control hardware. Depending on the sophistication of the FPGA, all the communication, motor control, signal processing, and other digital hardware can all be implemented with the programmable logic of the same FPGA, only limited by extent of programmable logic cells, memory blocks (RAM/ROM), and input-output blocks/IO translators. This means that with the right intellectual property (IP) blocks and design resources, the digital communication hardware, signal processing, motor control, and other functions can all be implemented with the same FPGA in real-time and deterministic hardware. This can also include advanced control algorithms, such as field-oriented control (FOC) algorithms for 3-phase permanent magnet synchronous motors (PMSM). Typically, FOC is very math-intensive and burdens an MCU/DSP substantially. Still, such an algorithm helps achieve high-performance levels from industrial servo motors that require precise torque control.

A benefit of FPGA motor/motion control is that multiple control systems can be implemented on an FPGA and operate in parallel in real-time. Software control systems implemented on MCUs/DSPs generally are processes in sequence. Though they can be processed rapidly, multiple cores can help with parallelism, though there are typically staggered control operations with these devices. MCU/DSP software development is generally considered less burdensome than FPGA programmable logic development. However, cost/power constraints are other key considerations for a given application (Figure 1).

 Figure 1: A high-level diagram of a field-oriented control algorithm implemented with an FPGA. (Source: Mouser Electronics)

Motor Driver Switch Technology Insights

The motor control signals from the motion/motor control, often speed, torque, position, and signals, along with current and/or voltage sensors included in the motor control boards or somewhere along the path to the motor, are converted by the motor drivers to higher power (voltage/current) version of the motor control signals. The motor driver circuits' task is to faithfully recreate the motor control signals at these higher power levels as efficiently and responsively as possible. 

For some motor types, such as PMSM and brushless direct current (BLDC) motors with 3-phases, a set of switches is needed for each phase. With multiple motion and motor driver circuits, modern AMRs require many switching circuits. All of which need to perform well for the AMR to function. Therefore, power density and reliability are also key characteristics of motor driver switches for industrial robotics. With more advanced stepper, servo, and BLDC/PMSM being used in industrial applications, AMR motor driver switches also need to be efficient at higher switching speeds and more compact to fit in the constrained spaces available for the AMR power electronics. The efficiency of motor driver switches directly impacts the size and complexity of the thermal management (heatsinks, etc.) needed for AMR power electronics. Smaller and more efficient motor driver switches can help reduce the overall footprint of AMR power electronics.

Two main motor driver technologies are currently widely available: silicon insulated-gate bipolar transistors (Si-IGBT) and silicon metal oxide semiconductor field effect transistors (Si-MOSFET). Si-IGBTs are legacy technology, though still widely applicable for lower-speed switching applications, where the high-speed and other features of Si-MOSFETs lend themselves better to newer and more complex motor control applications, such as industrial robotics. 

There is also an emerging technology, gallium nitride field effect transistors (GaN-FET), of which many varieties are beginning to be adopted in some aspects of industrial robotics. However, GaN-FET technology is relatively recent, and the drive circuitry is comparably complex and requires a very carefully controlled gate node stimulus. Hence, most modern industrial robotics are likely to be designed with Si-MOSFET technology, given the following benefits over legacy Si-IGBT solutions.

Though generally considered robust and cost-effective, Si-IGBTs are only efficient at relatively slow switching speeds and tend to exhibit high losses compared to MOSFETs. Given Si-IGBT's slow recovery characteristics, it isn't usually viable to use Si-IGBTs at higher frequencies beyond ~16kHz. Early silicon MOSFETs (Si-MOSFETs) didn't necessarily achieve the same power density levels as Si-IGBTs but did perform with lower switching losses at full-load conditions. A detracting characteristic of older Si-MOSFET technologies is that these devices tended to have poor internal body-diode recovery losses, which contribute to higher overall losses. However, Si-MOSFETs exhibit relatively linear current/voltage relationships during light-load operations, which can be advantageous.

The latest Si-MOSFETs feature significantly reduced drain-source on resistance (R_DS_ON), which minimizes conduction losses. Moreover, with these newer MOSFETs, the total gate charge (Q_G) is also improved, along with gate capacitances, further reducing overall driver losses. A soft recovery body diode (SRBD) in these MOSFET switches also reduces the reverse recovery (Q_RR) charge. With improved output charge (Q_OSS), light load efficiency tends to also be enhanced with newer Si-MOSFETs. An example of this is onsemi’s latest N-Channel MOSFET, the NTTFS012N10MD, which is made with onsemi’s advanced PowerTrench® process that also incorporates Shielded Gate technology.

Conclusion 

Autonomous mobile robotic (AMR) systems are taking over many industrial applications, including manufacturing, warehousing, and order fulfillment. Given that it is still necessary for human workers to be present alongside these robots, safety and reliability are paramount. Of AMR subsystems, motor control and drivers are vital elements that need to perform as designed to ensure a high level of safety and performance.