Acceleration Robotics (contact info) can help with this. We can leverage our experience in hardware acceleration and help create custom compute perception pipelines with GPUs and FPGAs that provide fast responses[1] while remaining deterministic.
Some benchmarks. ↩︎
