Case studies

Accelerating Robots in Industry

Accelerating metal part manufacturing with robots

Grupo Celsa, a leader in metal part manufacturing, faced challenges with slow, clunky robots powered by CPU-based vision proccesses, hindering production and quality control. Partnering with them, we accelerated their robot operations. By hardware accelerating their robotic vision pipelines by over 62x, we unlocked significant improvements in their robotic assemly processes. Robots can now operate faster, streamlining production.

Learn more

Accelerating industrial legacy robot manipulation

TEKNIA, an auto parts manufacturer, aimed to accelerate their robot production cycles. Through our collaboration, we first achieved a 16.21% increase in robot movement speed by optimizing trajectories, velocities, and accelerations. In a second stage, we reduced cycle time by 10.78% using double-manipulation techniques. These improvements empowered TEKNIA to produce an additional 800 units per day and over 20,000 units per month.

Learn more

Hardware Acceleration and Semiconductors for robots

FPGA IP core and GPU kernel designs and consulting

Foundation for High-Performance ROS 2 Communications

In partnership with AMD, we built the hardware groundwork for high-speed ROS 2 robot communications. Using FPGAs, we designed silicon building blocks for streamlined ROS 2 interactions. This involved a custom hardware networking stack (UDP/IP) working with an efficient Linux-bypassed DDS software layer. We also enhanced AMD's profiling tools to precisely track CPU-to-FPGA interactions, allowing system architects to fine-tune robot data flow and maximize performance.

Supercharging ROS with Agilex FPGAs and benchmarking

Our partnership with Intel enhanced Quartus suite and oneAPI frameworks, enabling them to seamlessly integrate with ROS. Through build system extensions and various connectors, system architects can now design custom accelerators for Intel's Agilex FPGAs, significantly accelerating robot operations within the familiar ROS environment. We also performed state-of-the art benchmarking to validate the performance gains and ensure peak efficiency.

Adaptive Computing Meets ROS

Xilinx partnership leveraged our FPGA mastery, optimizing ROS robot performance through custom IP cores and adaptive computing. Strategic intra-FPGA datapaths for ROS accelerated communications, and XEN hypervisor integration boosted performance and development efficiency within the familiar ROS environment.

Streamlining RISC-V FPGA SoCs

Our collaboration with Microchip optimized their RISC-V FPGA SoCs for ROS. Leveraging the FPGA fabric, we enabled a vendor-agnostic ROS development experience, mirroring the standard flow. This significantly boosted development speed, streamlined ROS workflows, and empowers next-generation robot architects.

Hardware Acceleration for Apex.AI

Our collaboration with Apex.AI supported their engineering team to leverage hardware acceleration. Through a series of consultations, we helped adapt Apex.Grace™ development kit architecture to embrace modern acceleration technologies, unlocking next-level performance on their safety-critical robotics software.

Hardware Acceleration Case Studies in Robotics