In the last issue, when talking about the development of MCU in the robot industry, it was mentioned that MCU is currently the most used in industrial robots and service robots. With every step of automation upgrade in the industrial field, the computing performance required by robots has also increased simultaneously. Various high-efficiency requirements, human-machine collaborative applications and various real-time computing require robots to provide precise motor control. Extensibility.

Obviously, FPGA manufacturers will not let MCUs relied on the dividends of motor control to reap the robot market. At present, the development trend of industrial robots is multi-axis and collaborative. This flexibility and expansion requirements are all for robots to perform a variety of sophisticated and complex anthropomorphic actions. Under this development trend, the number of motors and axes of industrial robots must be increased. When one of the axes rotates to a specific angle, the axis of the same linkage system must also rotate to the corresponding correct angle. Under such requirements, The coordination between the different systems has also become particularly important.

FPGAs can be applied to small nodes with decentralized control, or higher-performance FPGA SoCs can be applied to large nodes. The most obvious advantage of the system controlled by FPGA is that the driving response time will be greatly shortened, which will greatly improve the running speed of the entire system. The complex multi-axis computing requirements highlight the advantages of FPGA scalability and operation performance, and MCUs operating on a single node will be weak in dealing with such high performance requirements. Although DSP has higher processing efficiency, it still gives FPGA some breathing space in terms of scalability. How to make full use of the advantages of high performance and high scalability has become the key point for FPGA to break through in the robot market, which is a tripartite with MCU and DSP.

Robot FPGA high-efficiency operation control application

A design like this with full-scale robotic control can be seen from Xilinx's Versal AI Edge based on an Artix-7 FPGA and Zynq-7000 SoC design to see the benefits of this control approach.

From the perspective of Artix-7 FPGA alone, it provides a high-performance power consumption ratio structure (about 50%), transceiver line speed, DSP processing capability and AMS integration. 215K of logic cells and AXI IP and analog mixed-signal integration give the system a sufficiently high level of programmable integration.

The Zynq-7000 uses a single-core ARM Cortex-A9 processor paired with 28nm Artix-7 programmable logic, and provides a 6.25Gb/s transceiver, which can control costs in industrial applications such as motor control and embedded vision. It can also optimize system integration.

Versal AI Edge, a robot control based on the above components, can perform low-latency motor control and deterministic networking with metadata processing from visual and non-visual sensors to synchronize the entire system. For robots that require mobility, motion planning can be accelerated to meet navigation needs. This parallel processing of the control loop enables precise and deterministic control of the number of scalable motion axes, and has high scalability in multi-axis motion.

Not only Xilinx, but another FPGA manufacturer Altera also uses 28/20nm FPGA to enter the robot control market. Based on FPGA and FPGA SoC design, in addition to motion control, sensor bus management, camera bus management including HMI, etc. are integrated together. Reduce physical size and power consumption through high-level integration. Such control seems to better reflect the intelligence and adaptability of the robot.

Machine Vision FPGA Scalability Applications

FPGA with powerful parallel computing function can not only meet the needs of multi-axis robot motion, but also realize the flexible configuration of machine vision system with high flexibility. Lattice is such a strategy. Although lattice's FPGA also involves multi-axis motor control, its main application is on machine vision.

The inherent programmability of FPGAs enables them to support a variety of industrial communication protocols, making FPGAs a good choice for expanding communications. From the ECP5 processor board, its input and output boards can be combined and matched to connect to various image sensors and displays, providing maximum flexibility for interconnectivity.

write at the end

It is not difficult to see that FPGA has played a considerable role in the robotics industry by relying on high performance and scalability. Whether it is relying on high-performance processing power to seize the share of motor control from MCU, or relying on expansion to enter machine vision and DSP head-on confrontation, in the robot market, FPGA, MCU, and DSP can be said to be incompatible with each other, and undercurrents are surging. After all, no one wants to be crushed in this big market.

Responsible editor: haq

Leave a Reply

Your email address will not be published. Required fields are marked *