Support for RTC with async inference

[xfoukas]

According to this tutorial, async inference and RTC can be used together for maximum smoothness and reactivity. Is there some tutorial on how both can be enabled together? I was unable to find a way to enable RTC with the async_inference.robot_client.

[helper2424]

It's not implemented yet, it's the next step for RTC, after several fixes are merged.

[xianglunkai]

@helper2424
Excuse me, when RTC is applied to PI05 inference, enabling torch.compiler will result in a significant inference delay time during startup, causing the robotic arm to pause one by one.

[helper2424]

@xianglunkai I am not sure that compilation will work well, as RTC breaks computation graph. Probably its better to turn it off.

[xianglunkai]

@helper2424
Txs!