The path toward semiconductors for robots has two proper names whose mission is “to democratize the acceleration of” hardware in Robotics: Acceleration Robotics and Harvard University. start up Basque Country-based Semiconductor Robotics has joined forces with Harvard University professor Vijay Janapa Reddy and researchers from the Harvard Edge Computing Lab to present its latest advances in Japan next October. A paper on democratizing the use of data acceleration hardware In Robotics in a Scalable, Vendor- and Technology-Independent Way, under the title Robotcore: an open architecture for hardware acceleration in ROS 2The paper describes and reveals Reference implementation of architectural pillars and programming conventions needed to start acceleration hardware in robotics in a sustainable wayAvoiding lock-in of semiconductor suppliers.
In short, according to both the entities, the acceleration of hardware will enable the creation of custom computing architectures called accelerators or robot cores that will take advantage of computational parallelism, Thus, instead of relying solely on the Central Processing Unit (CPU), with an acceleration of hardware Via FPGA or GPU, roboticists can power robots faster with shorter computation times, lower power consumption and more deterministic responses, The central idea behind their research is to facilitate the process of using this technology, which, according to the researchers, with current solutions “requires experience in each platform. hardware specific(acceleration)”.
Open Architecture for Hardware Acceleration in ROS 2.
critical calculation acceleration
The research is developed under the leadership of Victor Meyerl-Vilches, a robotics specialist and former systems architect at Xilinx (now AMD), who left the company to form his own startup (Acceleration Robotics). Since then his work has focused on creating these robotic accelerators, or semiconductor building blocks, for robots that combine CPU, FPGA and GPU. a) yes, The resulting robotic accelerators deliver significant computation speeds compared to modern CPU performance, “Robots are deterministic machines,” says Meyer-Wilches, “so meeting deadlines is the most important feature in their calculations.”
“Their behaviors take the form of computational graphics, with data flowing between nodes via physical networks (communication buses), while being mapped to built-in sensors and actuators. The popular choice for creating robotic graphics these days is ROS. Most companies that make real robots use the framework of ROS or software Based on similar events, but they do it in a CPU-centric way. We demonstrate in our work that . how is the acceleration hardware The right combination of FPGA and GPU could revolutionize robotics, increase determinism and accelerate response times of robots allowing for new applications”, concludes the founder of Acceleration Robotics.
