For Companies
Joint research, contract research, academic consulting and lectures
Our lab works end to end, from integrated circuits (FPGA, ASIC, SRAM compute-in-memory) and edge AI to real-time robot control (imitation learning, acceleration of Vision-Language-Action models, low-latency bilateral control).
We are happy to discuss these topics with companies.
What we work on
See Research for details.
Design automation
AI for Circuits- Prototype flows that generate Verilog and layout (GDSII) from specifications or natural language
- Feasibility studies of introducing LLMs into an existing design flow
- Validation on open PDKs and FPGAs
AI hardware
Circuits for AI- Estimating and implementing speed, power and area of an AI model on FPGA/ASIC
- Applying our design assets: SRAM compute-in-memory macros measured on 22 nm silicon, a multiplication-free DNN accelerator
- Selecting and benchmarking AI chips and FPGAs
Robot control
Systems & Robotics- Retrofitting low-latency control (FPGA-based bilateral control, imitation learning) to existing robots
- Running and accelerating Vision-Language-Action models on edge devices
- Real-robot validation of teleoperation and imitation learning
Scope
- Silicon tape-out is costly and slow; we start with FPGA and simulation-based evaluation
- We do not do product development itself (mass-production design, maintenance); the work needs a research component
- The lab has no GPU cluster or large measurement facilities; if these are required, let us talk about how to proceed
How we collaborate
- Joint research
- Contract research (problem solving first; deliverable is a report; publication is optional)
- Academic consulting (selecting and evaluating AI chips and FPGAs, introducing design automation, robot control)
- Lectures and corporate training (e.g., “Introduction to AI Chips: Hardware Acceleration for Machine Learning”)
Selected results
- SRAM compute-in-memory macros measured on 22 nm silicon (IEEE TVLSI 2026, ICCAD 2026)
- Multiplication-free all-digital DNN accelerator (DAC 2025)
- Low-latency bilateral control on FPGA and accelerated motion generation for Vision-Language-Action models (RSJ 2026)
- Uncertainty-aware haptic shared control for humanoid teleoperation (IEEE RA-L 2023)
- Low-latency spiking-neural-network inference applied to robot imitation learning (Scientific Reports 2026)
See Publications for the full list.
Contact
If any of this sounds relevant, feel free to get in touch — an exploratory conversation is perfectly fine. We can also talk under an NDA.
-
Laboratory: pj.u-ayogan ta onawa
-
Nagoya University collaboration office: inquiry form