Component plugins

All components are plugins (see flapsim plugins). They exchange named signals: a sensor publishes imu/gyro, a controller reads it and publishes cmd/stroke/frequency_hz, the stroke actuator consumes the command. Plugins run in insertion order within each phase (pre_control for sensors, control for controllers, actuate for actuators, forces for aero).

Actuators

Plugin	What it does	Commands	Signals
actuator/sinusoidal_stroke	Kinematic q = d A sin(2 pi f t) on stroke joints (kinematic writes velocities, position_only does not)	frequency_hz, amplitude_rad	frequency_hz, amplitude_rad, phase
actuator/pd_position	Gravity-compensated PD with torque limit, or kinematic hold; joints carry a role label	<joint> target	<joint>
actuator/servo	pd_position with a rate-limited, first-order lagged target	<joint> desired	<joint>, <joint>/desired
actuator/dc_motor_crank	DC motor (i = (V - k_e w)/R, tau = k_t i), gearbox N, crank q = d A sin(theta_c); loaded by the aero torque about the stroke joints	voltage_v	speed_rad_s, frequency_hz, current_a, torque_load, power_w, ...

Sensors

Every sensor samples at rate_hz through a Conditioner: first-order low-pass tau, pure delay delay_s, Gaussian noise_std, a bias drawn per episode from bias_range, and quantisation resolution.

Plugin	Measures	Defaults
sensor/imu	euler (low-passed attitude), gyro (body rates), accel (specific force R^T (a - g)), quat	500 Hz, 20 ms attitude lag, 0.02 deg / 0.002 rad/s / 0.02 m/s^2 noise, 0.01 m/s^2 bias
sensor/rangefinder	slant range along a body axis to the ground plane or by ray cast, range_rate	500 Hz, 20 ms delay, 5 mm noise, 4 m range
sensor/encoder	joint position (quantised by counts_per_rev) and velocity	continuous
sensor/wrench	interaction wrench with the parent body (cfrc_int): a load cell under a fixed-base robot	averaging average_s

Controllers

Plugin	Role
controller/hover_pid	Period-averaged altitude (frequency) and attitude (gimbal) loops with waypoints
controller/schedule	Time tables of set points ([[t, value], ...]), stepped or interpolated
controller/callback	Any Python function callback(sim, controller) -> {command: value} at a rate

Recording and visualisation

• recorder/memory: samples glob-matched signals every every seconds; to_dict, to_npz, to_csv, to_hdf5.
• flapsim.io.ForceArrows: viewer or renderer overlay with each wing's force arrow at its centre of pressure (flapsim run ... --overlay forces).
• flapsim.io.record_video / flapsim record: offscreen rendering to GIF, PNG frames or MP4 (MUJOCO_GL=osmesa or egl on headless machines).