Control Delay
The time lag between issuing a control command and the actuator actually responding. Introducing random delays during robot training can enhance the robustness of policies to the real environment.
Timeline (last 90 days)
The developers replaced the idealized actuator with a DC motor torque-speed model, added random control delays of up to 25 milliseconds, and applied external pushes during training.
Introducing random control delays of up to 25 milliseconds during training can enhance the policy's robustness to delays and improve stability in real environments.
All Facts (2)
The developers replaced the idealized actuator with a DC motor torque-speed model, added random control delays of up to 25 milliseconds, and applied external pushes during training.
50%UnverifiedIntroducing random control delays of up to 25 milliseconds during training can enhance the policy's robustness to delays and improve stability in real environments.
50%