Systems, Boards and Chips
Embedded electronics solutions for robotic systems continue to advance, as chip-, board- and system-level solutions evolve to meet new demands. Real-time motion control, machine vision and artificial intelligence are all part of the mix.
It’s hard to generalize when it comes to the broad application segment of robotics. The needs of factory robots tasked to do 24/7 production floor duties are different than those of a consumer or commercial robot service robot. That said, the fundamental elements are similar. Robotics systems need intelligent, real-time sensing and control, precision management of all kinds of motors and an ability to leverage embedded processing at artificial intelligent (AI) and machine learning levels (ML) of performance.
To meet the diverse requirements of robotic system development, technology providers across a wide spectrum continue to churn out innovative solutions. These solutions range from highly integrated box-level systems, to advanced development kits to board- and chip level solutions for sensing and motion control. Over the past 12 months, a new crop of these products has emerged all meeting critical challenges in robotics design.
OS FOR ROBOTICS
Now, there’s even an operating system aimed specifically at robotics applications. Provided by the Open Source Robotics Foundation, the ROS (Robot Operating System) is a software development kit that helps developers build robot applications. ROS is all open source, and the kit includes drivers, state-of-the-art algorithms and powerful developer tools. ROS version 2 (ROS 2) was launched in June.
Leveraging those resources, in June ADLINK Technology launched two box-level systems that run the ROS 2 operating system (Figure 1). The first is the “ROScube-I with Intel”, serving as a real-time ROS 2 robot controller for advanced robotic applications. The box-level system is based on Intel Xeon E, 9th Gen Intel Core i7/i3 and 8th Gen Intel Core i5 processors. Aside from rich I/O connectivity, the unit supports a wide variety of sensors and actuators to meet the needs of a wide range of robotic applications. The ROScube-I supports an extension box for convenient functional and performance expansion with Intel VPU cards and the Intel Distribution of OpenVINO toolkit for computation of AI algorithms and inference.
(a) ROScube-I, a real-time ROS 2 robot controller based on Intel Xeon E, 9th Gen Intel Core i7/i3 and 8th Gen Intel Core i5 processors. (b) ROScube-X Series, a real-time ROS 2 controller based on NVIDIA’s Jetson AGX Xavier module, which has an integrated NVIDIA Volta GPU.
The ROScube-I provides comprehensive I/O for connecting a wide range of devices, including real-time I/O for environmental sensors (CAN, GPIO and COM). Real-time middleware is provided for communication between software components and devices. A hypervisor ensures safe mission critical mission execution. The ruggedized systems feature secure connectivity with locking USB ports. The system provides optional support for RTOSes such as Wind River’s VxWorks.
The second ROS 2 offering from ADLINK is its ROScube-X Series, also a real-time Open Robotics ROS 2 controller for autonomous robotics applications. But this box-level system embeds NVIDIA’s Jetson AGX Xavier module, which has an integrated NVIDIA Volta GPU, dual deep-learning accelerators and a wide variety of interfaces, including GMSL2 camera connectors for advanced robotics system integration. The ROScube-X supports the full complement of resources provided by the NVIDIA JetPack SDK. The system is specifically suited for robotics applications demanding high-performance AI computing capabilities with minimal power consumption, says the company.
The ROScube-X’s GPU processing provides powerful artificial intelligence (AI) computing for autonomous robotics development. The system is qualified with Amazon Web Services (AWS) Internet of Things (IoT) solutions and works with AWS RoboMaker. ROS/ROS 2 with Eclipse Cyclone DDS has been tested and optimized on the system. The unit’s performance per watt with power consumption clocks as low as 20W. Like the ROScube-I, the ROScube-X boasts rugged construction and secure connectivity with locking USB ports. It likewise provides comprehensive I/O for a wide range of devices along with time synchronization for eight GMSL2 cameras and inertial measurement unit (IMU). Optional CANbus and LTE-A modules are available.
There’s no doubt we’re transitioning into a 5G world, and robotics is feeling its influence. Along those lines, in June Qualcomm announced its Qualcomm Robotics RB5 platform, a system designed specifically for robotics. The product follows its previous Robotics RB3 platform, which has enjoyed broad adoption in a wide array of robotics and drone products, says the company.
The Qualcomm Robotics RB5 platform is comprised of a set of hardware, software and development tools. The Qualcomm Robotics RB5 platform was designed to meld Qualcomm’s expertise in 5G and AI to empower developers and manufacturers to create the next generation of high-compute, low-power robots and drones (Figure 2).
The Qualcomm Robotics RB5 platform was designed to meld Qualcomm’s expertise in 5G and AI to empower developers and manufacturers to create the next generation of high-compute, low-power robots. (a) Robotics RB5 Platform Core; (b) Vision Kit version.
The platform’s Qualcomm QRB5165 processor, customized for robotics applications, offers a heterogeneous computing architecture coupled with the 5th-gen Qualcomm AI Engine, delivering 15 tera operations per second (TOPS) of AI performance for running complex AI and deep learning workloads. The processor also ML inferencing at the edge under restricted power budgets using the new Qualcomm Hexagon Tensor Accelerator (HTA), an image signal processor (ISP) with support for seven concurrent cameras and a dedicated computer vision engine for enhanced video analytics (EVA).
With support for 4G and 5G connectivity speeds via a companion module, the Qualcomm Robotics RB5 platform is expected to pave the way for the proliferation of 5G in robotics and intelligent systems. The solution is available in multiple options, including commercial and industrial-grade temperature ranges and an option for extended lifecycle until 2029.
Software support for the RB5 Development Kit includes Linux, Ubuntu and ROS 2, as well as pre-integrated drivers for various cameras, sensors and 5G connectivity. It also provides support for OpenCL, OpenGLES and OpenCV. The development kit includes support for the Intel RealSense Depth Camera D435i and Panasonic TOF Camera to provide depth-sensing capabilities. Thanks to Qualcomm’s strategic collaboration with TDK, the kit integrates TDK’s 6-axis, high performance ICM-42688-P IMU, accompanied by an ICP-10111 barometric pressure and a T5818 Digital (PDM) bottom port microphone. TDK also enables the development kit with its ultrasonic Time-Of-Flight CH101 and CH201 solutions, in addition to its embedded motor control (HVC4223F) and high performance IMU with built-in redundancy (IIM-46220).
VISION AND VMS
Vision is a key component of many robotics systems. Supporting those needs, in June Congatec rolled out its Intel IoT RFP (Ready For Production) Kit. Congatec positions it as a workload consolidation kit for vision based situational awareness that is qualified by Intel. Based on a COM Express Type 6 module equipped with Intel Xeon E2 processor, the Intel IoT RFP kit has three virtual machines (VMs) built on Real-Time Systems’ hypervisor technology for workload consolidation in vision applications (Figure 3).
Based on a COM Express Type 6 module equipped with Intel Xeon E2 processor, the Intel IoT RFP kit has three VMs built on Real-Time Systems’ hypervisor technology for workload consolidation in vision applications.
The Congatec kit targets the next generation of vision based collaborative robotics, automation controls and autonomous vehicles that have to tackle multiple tasks in parallel, including situational awareness utilizing deep learning-based AI algorithms. The Real-Time Systems-based VMs make it possible to consolidate the different tasks on a single edge computing platform, which ultimately saves costs. The Intel OpenVino software delivers the appropriate AI for situational awareness. According to Congatec, OEMs just need to load their control programs to the real-time VM and they are ready to enrich their real-time control with data from the situational awareness VM and to communicate in real-time with IIoT/Industry 4.0 counterparts to enable tactile Internet controls.
Aside from the COM Express module, the Intel IoT RFP Kit includes a Basler vision camera, a pendulum controlled by a demo controller and an Intel Arria 10 FPGA card from REFLEX CES. The platform has three pre-installed application ready VMs based on Real-Time Systems’ hypervisor technology. One VM analyses videos on the basis of Intel OpenVino software, the second VM runs real-time Linux to control the balance of an inverted pendulum in real-time. The third partition hosts a gateway for IIoT/Industry 4.0 connectivity.
PRECISE MOTOR CONTROL
Motors are used in all sorts of embedded systems, but robot designs have their own specific requirements when it comes to motor control. With that in mind, last Fall STMicroelectronics (ST) announced a coloration with precision motor vendor Maxon. At the SPS Germany 2019 trade show, the two firms jointly demoed a servo control kit they developed. Called EVALKIT-ROBOT-1, the kit is a plug-and-play solution aimed to help users easily approach the world of precise positioning and high-end motion in servo drives and robotics (Figure 4). A Maxon 100W BLDC (brushless) motor with built-in 1024-pulse incremental encoder is included in the kit. The solution ensures smoothness and balance to allow fine control even at low rotor speeds.
The EVALKIT-ROBOT-1 kit is a plug-and-play solution aimed to help users easily approach the world of precise positioning and high-end motion in servo drives and robotics. A Maxon 100W BLDC motor with built in 1024-pulse incremental encoder is included in the kit.
The servo control board supplied with the kit contains ST’s STSPIN32F0A intelligent 3-phase motor controller and a complete inverter stage built with ST power transistors ready to connect to the motor. Motor-control firmware is also included, making it easy for users to start the motor and begin sending commands.
ST’s STSPIN32F0A system-in package (SiP) contains critical circuitry for motor control, including an STM32F031C6 microcontoller (MCU) and three-phase inverter driver in a compact 7mm × 7mm VFQFPN package. The MCU comes loaded with plug-and-play firmware for MODBUS communication and field-oriented control (FOC) with precise positioning capabilities. Power management and current sensing circuitry are also embedded in the device making it more flexible and versatile.
The Maxon EC-i 40, 40mm-diameter, 100W BLDC motor embeds a Maxon ENX 16 EASY 1024-pulse incremental encoder for precision control. Hall-effect sensors for detecting rotor position are included. The motor features an optimized design for high output torque with low cogging torque, which permits smooth motion across the speed range and enhances positioning precision.
HATs FOR ROBOTICS
The popularity of Raspberry Pi keeps ramping up. And, as a wider variety of Raspberry Pi add-ons in form of HATs (hardware attached on top) emerge, there are more and more applications that can easily embed Raspberry Pi computing. Serving robotics needs, in April, Sparkfun launched two pHATs for the Raspberry Pi, Jetson Nano and Google Coral. The Auto pHAT offers robotics motor and servo controls and an IMU while the Top pHAT has a 2.3″ color display plus LEDs, buttons, mic and speaker for ML prototyping (Figure 5).
Designed for use with Raspberry Pi SBCs, the Auto pHAT (a) offers robotics motor and servo controls and an IMU. The Top pHAT (b) has a 2.3″ color display plus LEDs, buttons, mic and speaker for machine learning (ML) prototyping.
Both pHATs support the NVIDIA Jetson Nano and Google Coral Dev Board in addition to any Raspberry Pi with a 40-pin GPIO connector. For its part, Sparkfun follows its pHAT (which stands for partial HAT) approach, which doesn’t support all the specs of the official Raspberry Pi HAT. That said, both of the new pHAT products are open-source hardware modules with posted schematics and Eagle files.
The SparkFun Auto pHAT can drive two small DC motors with or without encoders and can support up to four servo motors via I2C. Python packages are available for major components. The Auto pHAT’s servo control is based on the PWM (pulse width modulation)-enabled SparkFun Servo pHAT, which has an I2C link that frees up Raspberry Pi GPIO pins for other purposes. There is also a Qwiic connector for interfacing with the I2C bus found on Qwiic modules.
The SparkFun Top pHAT is a UI and control panel for Raspberry Pi-based systems. The pHAT sits on top of the Pi or stacked HATs. It lacks the 40-pin passthrough, but offers a 2.3″ color TFT display, buttons, LEDs and audio UI. Applications include ML prototyping, voice control, camera control feedback and system status display.
The Top pHAT is equipped with dual user buttons and a programmable joystick for GUI menus and robot control. Other features include a Qwiic connector, 6x LEDs and an off switch. Voice control is supported with 2x microphones and a mono speaker. Just like the Auto pHAT, the Top pHAT does not inhibit access to the Raspberry Pi’s camera or display connectors.
TINY SERVO DRIVES
Innovations continue in the area of servo drives, mostly by companies specializing in the technology. One vector these designs are taking is a reduction in size, which serves the needs of robotics designs. Along just those lines, in February, Advanced Motion Controls made three new additions to its FlexPro servo drive family, the FM series. Designed with compact form factor and power density in mind, the micro-sized FM drives can outperform larger-sized digital servo drives and still be integrated into tight spaces, according to the company (Figure 6). The FM drives are equipped with an additional interface board that makes it easy to integrate into your system with standard connectors.
Designed with compact form factor and power density in mind, the micro-sized 50.8mm × 25.4mm × 26mm FM drives can outperform larger-sized digital servo drives and still be integrated into tight spaces. The FM drives are equipped with an additional interface board that makes it easy to integrate into your system with standard connectors.
FM060-5-EM and FM060-10-EM are connected to the power supply and motor using screw terminals, while the FM060-25-EM features pre-soldered connection leads. The units measure 50.8mm × 25.4mm × 26 mm (2″ × 1″ × 1.03″) with 10VDC to 55VDC input. Current specs are 5A continuous, 10A peak for the FM060-5-EM, 10A continuous, 20A peak for the FM060-10-EM and 25A continuous, 50A peak for the FM060-25-EM. The devices feature incremental encoder and BISS C-mode feedback. They operate in torque, velocity and position modes. The small size of the FM drives make them well suited for cobots (collaborative robots) and autonomous ground vehicles (AGVs).
The FlexPro devices feature the IMPACT architecture. It allows the servo drives to maintain the intelligence and power of larger-sized servo drives while fitting into much more compact packages. IMPACT (Integrated Motion Platform And Control Technology) architecture involves the stacking of circuit boards with creative selection and placement of high-power components. This allows for much higher power density than previously produced servo drives. IMPACT makes its debut with the FE060-25-EM. The company says this technology will be used in subsequent FlexPro drives as well as any future custom products where it is desired.
INDUCTIVE POSITION SENSING
Industrial robotics systems require extreme precision control, and that means position sensing that’s extremely accurate. In an example solution to such needs, in June Renesas Electronics introduced the magnet-free IPS2200 inductive position sensor. Featuring high accuracy and speed, total stray field immunity and efficient motor integration in a thin and lightweight form factor, the IPS2200 is well suited for use as an absolute position sensor in a wide range of industrial, medical and robot applications (Figure 7). The sensor allows engineers to cost effectively tailor sensor design for their applications and maximize the performance of the sensor’s accuracy.
Featuring high accuracy and speed, total stray field immunity and efficient motor integration in a thin and lightweight form factor, the IPS2200 is well suited for use as an absolute position sensor in a wide range of robot applications. The sensor allows robot designers to tailor sensor design to their applications and maximize the performance of the sensor’s accuracy.
Inductive position sensing is changing the game for industrial motor commutation as demands for high accuracy, high efficiency and cost efficiency increase, particularly for multi-pole pair motors and off-axis applications, says Renesas. The IPS2200 is designed around the motor, allowing system developers to match the number of sectors to the pole pairs of the motor to maximize accuracy, accommodating both off-axis (through shaft and side shaft) and on-axis positioning.
THINNER AND LIGHTER
The magnet-free IPS2200 is up to 10x thinner and up to 100x lighter with up to 250kRPM electrical speeds compared with traditional resolvers. The sensor’s thin and light form factor and total stray field immunity enables easier motor integration and provides the standard materials required for developers to manufacture their own resolver replacement—reducing bill of materials costs. With its four- or six-wire operation, the IPS2200 provides up to 10x faster speeds and very low latency compared with resolver- or magnetic-based solutions.
The IPS2200 chips are industrial qualified, with stable operation in harsh environments and -40°C to +125°C ambient temperatures. Interfaces are sin/cos single ended or differential. Voltage supplies are 3.3V ±10% or 5.0V ±10% and rotational speed are supported up to 250.000RPM (electrical). Support is included for sin/cos gain mismatch and offset compensation along with overvoltage, reverse polarity and short-circuit protections. Digital programming can be done using the device’s I²C or SPI interface.
Time-of-Flight (ToF) sensors play a critical role in many types of robotic systems, particularly systems like service robots and robotic vacuum cleaners. With their high accuracy and fast response times, ToF sensors also enhance the performance of all devices requiring precise movement control. In May, STMicroelectronics (ST), extended the capabilities of its FlightSense ToF ranging sensors by introducing the VL53L3CX with patented histogram algorithms that allow measuring distances to multiple objects as well as increasing accuracy (Figure 8).
The VL53L3CX ToF sensors feature patented histogram algorithms that allow measuring distances to multiple objects as well as increasing accuracy. The VL53L3CX measures object ranges from 2.5cm to 3m, unaffected by the target color or reflectance.
Unlike conventional infrared sensors, the VL53L3CX measures object ranges from 2.5cm to 3m, unaffected by the target color or reflectance. This allows designers to introduce powerful new features to their products, such as enabling occupancy detectors to provide error-free sensing by ignoring unwanted background or foreground objects, or reporting the exact distances to multiple targets within the sensor’s field-of-view.
The ST patented histogram algorithms increase cover-glass crosstalk immunity and allow real-time smudge compensation preventing external contamination from adversely affecting the ranging accuracy of, for example, vacuum cleaners or equipment that may be used in a dusty industrial environment. Ranging under ambient lighting is also improved.
In addition, the VL53L3CX has superior linearity that increases short-distance measurement accuracy, enhancing wall tracking, faster cliff detection, and obstacle avoidance in equipment such as service robots and vacuum cleaners—markets in which ST has already enjoyed considerable commercial success. Like all FlightSense sensors, the VL53L3CX features a compact, all-in-one package design that eases integration into your devices, as well as low power consumption that helps extend battery runtime.
MCU FOR INDUSTRIAL ROBOTICS
While application-specific MCUs are nothing new, few are aimed specifically at robotics. One problem for industrial robotics is the integrated networking needed for MCU solutions. Along just those lines, in April Renesas Electronics announced its RX72N Group and RX66N Group of 32-bit MCUs. The chips are new additions to the RX Family that combine equipment control and networking functions on a single chip. Built around the Renesas proprietary RXv3 CPU core, the RX72N features a maximum operating frequency of 240MHz and two Ethernet channels, and the RX66N features a maximum operating frequency of 120MHz and one Ethernet channel.
In the industrial equipment market, performance and functionality improvements lead to larger program code sizes, says Renesas. As a result, storage capacity and read speed play key roles in determining real-time performance. The RX72N and RX66N devices offer up to 4MB of on-chip flash memory 1MB of SRAM.
With this combination, the MCUs can implement single-chip equipment control and networking function, enhancing the real-time performance of applications such as industrial robots, general-purpose inverters, programmable logic controllers (PLCs) and remote I/O devices. They also make it possible to improve the exterior design and operability of a wide range of industrial automation gear.
ADLINK Technology | www.adlinktech.com
Advanced Motion Controls | www.a-m-c.com
Congatec | www.congatec.com
Maxon | www.maxongroup.com
Qualcomm | www.qualcomm.com
Renesas Electronics | www.renesas.com
Sparkfun | www.sparkfun.com
STMicroelectronics | www.st.com
PUBLISHED IN CIRCUIT CELLAR MAGAZINE • SEPTEMBER 2020 #362 – Get a PDF of the issueSponsor this Article
Jeff served as Editor-in-Chief for both LinuxGizmos.com and its sister publication, Circuit Cellar magazine 6/2017—3/2022. In nearly three decades of covering the embedded electronics and computing industry, Jeff has also held senior editorial positions at EE Times, Computer Design, Electronic Design, Embedded Systems Development, and COTS Journal. His knowledge spans a broad range of electronics and computing topics, including CPUs, MCUs, memory, storage, graphics, power supplies, software development, and real-time OSes.