closed-loop control
A framework drawing on the sensor-actuator model from cybernetics, in which the agent reads value signals from hidden states at each step and adjusts model preferences in real time, forming a feedback loop. It is used to establish causal evidence linking representations to decision-making.
Timeline (last 90 days)
The closed-loop experiments provide behavioral-level evidence that "representations can be utilized" and validate the causal chain: if value readout were unrelated to decision-making, using it to drive interventions should not bring systematic improvement.
Common gait planning methods include open-loop gaits based on preset motion sequences and closed-loop control based on sensor feedback.
Closed-loop control refers to a system adjusting the control input based on real-time output feedback from the controlled object, rather than merely executing a preset program.
Closed-loop control introduces IMU, foot contact switch, or encoder feedback, and adjusts the gait phase through PID or a state machine.
If sensor noise reaches tens of nanometers, the controller of a sub-micrometer closed-loop system may treat the noise as a real deviation and adjust repeatedly, thereby introducing jitter.
Closed-loop control can compensate for uncertain factors such as mechanical transmission backlash, thermal drift, and load disturbances, whereas open-loop systems cannot.
The heartbeat cycle is about 0.8 seconds, and if the control loop latency exceeds a hundred milliseconds, the robot will be carried away from the target area by blood flow.
Any delay, noise, or calibration error in any step of the closed-loop control process can cause the robot to fail to react in time during dynamic movements and fall.
Dancing is essentially open-loop execution of preset trajectories, which is fundamentally different from closed-loop tasks requiring real-time perception and decision-making (such as grasping, obstacle avoidance, and human-robot collaboration).
All Facts (9)
The closed-loop experiments provide behavioral-level evidence that "representations can be utilized" and validate the causal chain: if value readout were unrelated to decision-making, using it to drive interventions should not bring systematic improvement.
50%UnverifiedCommon gait planning methods include open-loop gaits based on preset motion sequences and closed-loop control based on sensor feedback.
50%UnverifiedClosed-loop control refers to a system adjusting the control input based on real-time output feedback from the controlled object, rather than merely executing a preset program.
50%UnverifiedClosed-loop control introduces IMU, foot contact switch, or encoder feedback, and adjusts the gait phase through PID or a state machine.
50%UnverifiedIf sensor noise reaches tens of nanometers, the controller of a sub-micrometer closed-loop system may treat the noise as a real deviation and adjust repeatedly, thereby introducing jitter.
50%UnverifiedClosed-loop control can compensate for uncertain factors such as mechanical transmission backlash, thermal drift, and load disturbances, whereas open-loop systems cannot.
50%UnverifiedThe heartbeat cycle is about 0.8 seconds, and if the control loop latency exceeds a hundred milliseconds, the robot will be carried away from the target area by blood flow.
50%UnverifiedAny delay, noise, or calibration error in any step of the closed-loop control process can cause the robot to fail to react in time during dynamic movements and fall.
50%UnverifiedDancing is essentially open-loop execution of preset trajectories, which is fundamentally different from closed-loop tasks requiring real-time perception and decision-making (such as grasping, obstacle avoidance, and human-robot collaboration).
50%