Webots Sensors Skill

Use this skill to implement, configure, and debug sensor pipelines in Webots controllers. Focus on device acquisition, sampling-period control, data extraction, and sensor-specific post-processing. Exclude actuator control and basic controller bootstrap logic.

When to use

Activate this skill when tasks involve any of the following:

• Camera
• Lidar
• DistanceSensor
• GPS
• Gyro
• Accelerometer
• InertialUnit
• Compass
• TouchSensor
• RangeFinder
• Radar
• LightSensor
• PositionSensor
• Receiver
• Altimeter

Core execution pattern

Apply this sequence for all Webots sensors:

1. Acquire device by DEF name.
2. Enable sensor with a sampling period (typically controller timestep).
3. Advance simulation with step.
4. Read sensor value(s) with the appropriate getter.
5. Process, filter, or fuse data.

Python universal pattern:

sensor = robot.getDevice("sensor_name")
sensor.enable(timestep)  # MUST enable before reading
while robot.step(timestep) != -1:
    value = sensor.getValue()  # or getValues() for vector sensors

C universal pattern:

WbDeviceTag sensor = wb_robot_get_device("sensor_name");
wb_distance_sensor_enable(sensor, time_step); /* use sensor-specific enable */
while (wb_robot_step(time_step) != -1) {
  double value = wb_distance_sensor_get_value(sensor); /* or vector getter */
}

Sampling and performance rules

• Match sensor sampling period to controller timestep unless slower sampling is acceptable.
• Increase sampling period for expensive devices (Camera, Lidar, RangeFinder) to reduce CPU load.
• Pull only required outputs (for example, avoid point cloud retrieval when only scalar ranges are needed).
• Keep image/point-cloud conversions out of tight loops when possible.

Sensor playbooks (Python + C)

Camera

Use for RGB/RGBA image acquisition and optional object recognition.

• Key operations: enable, getImage, getWidth, getHeight, imageGetRed/Green/Blue/Gray (language-dependent helpers).
• Recognition: enable with recognitionEnable, then call getRecognitionObjects.
• Effects and optics from node fields: noise, motion blur, lens distortion, depth of field, fish-eye, spherical projection.
• C raw image API (wb_camera_get_image) returns byte buffer; convert to array/matrix for processing.

Python example:

camera = robot.getDevice("front_camera")
camera.enable(timestep)
camera.recognitionEnable(timestep)

width = camera.getWidth()
height = camera.getHeight()

while robot.step(timestep) != -1:
    image = camera.getImage()  # bytes in BGRA in Webots Python API
    # Simple obstacle cue: average center brightness drop
    center_x = width // 2
    center_y = height // 2
    gray = camera.imageGetGray(image, width, center_x, center_y)

    objects = camera.getRecognitionObjects()
    if len(objects) > 0 or gray < 40:
        obstacle_detected = True
    else:
        obstacle_detected = False

C example:

WbDeviceTag cam = wb_robot_get_device("front_camera");
wb_camera_enable(cam, time_step);
wb_camera_recognition_enable(cam, time_step);

int width = wb_camera_get_width(cam);
int height = wb_camera_get_height(cam);

while (wb_robot_step(time_step) != -1) {
  const unsigned char *img = wb_camera_get_image(cam);
  int cx = width / 2;
  int cy = height / 2;
  int gray = wb_camera_image_get_gray(img, width, cx, cy);

  int count = wb_camera_recognition_get_number_of_objects(cam);
  if (count > 0 || gray < 40) {
    /* obstacle_detected = true */
  }
}

Lidar

Use for depth scans and optional 3D point cloud.

• getRangeImage() provides distances.
• getPointCloud() provides LidarPoint entries with x, y, z, layer_id, time.
• Important fields: horizontalResolution, numberOfLayers, fieldOfView, minRange, maxRange.

Python example (360-degree scan processing):

lidar = robot.getDevice("lidar")
lidar.enable(timestep)
lidar.enablePointCloud()

while robot.step(timestep) != -1:
    ranges = lidar.getRangeImage()
    min_distance = min(ranges)
    points = lidar.getPointCloud()  # list of LidarPoint
    near_points = [p for p in points if p.x * p.x + p.y * p.y < 0.25]

C example:

WbDeviceTag lidar = wb_robot_get_device("lidar");
wb_lidar_enable(lidar, time_step);
wb_lidar_enable_point_cloud(lidar);

while (wb_robot_step(time_step) != -1) {
  const float *ranges = wb_lidar_get_range_image(lidar);
  int n = wb_lidar_get_horizontal_resolution(lidar) * wb_lidar_get_number_of_layers(lidar);
  float min_d = ranges[0];
  for (int i = 1; i < n; ++i)
    if (ranges[i] < min_d) min_d = ranges[i];

  const WbLidarPoint *cloud = wb_lidar_get_point_cloud(lidar);
  (void)cloud;
}

DistanceSensor

Use for short-range proximity with infra-red, sonar, or laser models.

• getValue() returns raw sensor reading.
• Convert raw reading to distance via node lookupTable.
• Sensor type field controls infra-red/sonar/laser behavior.

Python example (IR wall following):

left_ir = robot.getDevice("left_ir")
right_ir = robot.getDevice("right_ir")
left_ir.enable(timestep)
right_ir.enable(timestep)

while robot.step(timestep) != -1:
    l = left_ir.getValue()
    r = right_ir.getValue()
    error = l - r
    wall_on_left = l > 80.0

C example:

WbDeviceTag left_ir = wb_robot_get_device("left_ir");
WbDeviceTag right_ir = wb_robot_get_device("right_ir");
wb_distance_sensor_enable(left_ir, time_step);
wb_distance_sensor_enable(right_ir, time_step);

while (wb_robot_step(time_step) != -1) {
  double l = wb_distance_sensor_get_value(left_ir);
  double r = wb_distance_sensor_get_value(right_ir);
  double error = l - r;
  (void)error;
}

GPS

• getValues() returns [x, y, z].
• getSpeed() returns scalar speed.
• getSpeedVector() returns [vx, vy, vz].
• coordinateSystem: "local" or "WGS84" at node level.

Python:

gps = robot.getDevice("gps")
gps.enable(timestep)

while robot.step(timestep) != -1:
    x, y, z = gps.getValues()
    speed = gps.getSpeed()
    vx, vy, vz = gps.getSpeedVector()

C:

WbDeviceTag gps = wb_robot_get_device("gps");
wb_gps_enable(gps, time_step);

while (wb_robot_step(time_step) != -1) {
  const double *p = wb_gps_get_values(gps);
  double speed = wb_gps_get_speed(gps);
  const double *v = wb_gps_get_speed_vector(gps);
  (void)p; (void)speed; (void)v;
}

Gyro

• getValues() returns angular velocity [wx, wy, wz] in rad/s.

Python:

gyro = robot.getDevice("gyro")
gyro.enable(timestep)

while robot.step(timestep) != -1:
    wx, wy, wz = gyro.getValues()

C:

WbDeviceTag gyro = wb_robot_get_device("gyro");
wb_gyro_enable(gyro, time_step);

while (wb_robot_step(time_step) != -1) {
  const double *w = wb_gyro_get_values(gyro);
  (void)w;
}

Accelerometer

• getValues() returns [ax, ay, az] in m/s^2, including gravity.

Python:

acc = robot.getDevice("accelerometer")
acc.enable(timestep)

while robot.step(timestep) != -1:
    ax, ay, az = acc.getValues()

C:

WbDeviceTag acc = wb_robot_get_device("accelerometer");
wb_accelerometer_enable(acc, time_step);

while (wb_robot_step(time_step) != -1) {
  const double *a = wb_accelerometer_get_values(acc);
  (void)a;
}

InertialUnit

• getRollPitchYaw() returns [roll, pitch, yaw] in radians.
• getQuaternion() returns [w, x, y, z].

Python:

imu = robot.getDevice("inertial unit")
imu.enable(timestep)

while robot.step(timestep) != -1:
    roll, pitch, yaw = imu.getRollPitchYaw()
    w, x, y, z = imu.getQuaternion()

C:

WbDeviceTag imu = wb_robot_get_device("inertial unit");
wb_inertial_unit_enable(imu, time_step);

while (wb_robot_step(time_step) != -1) {
  const double *rpy = wb_inertial_unit_get_roll_pitch_yaw(imu);
  const double *q = wb_inertial_unit_get_quaternion(imu);
  (void)rpy; (void)q;
}

Compass

• getValues() returns normalized 3D vector to magnetic north.

Python:

compass = robot.getDevice("compass")
compass.enable(timestep)

while robot.step(timestep) != -1:
    nx, ny, nz = compass.getValues()

C:

WbDeviceTag compass = wb_robot_get_device("compass");
wb_compass_enable(compass, time_step);

while (wb_robot_step(time_step) != -1) {
  const double *north = wb_compass_get_values(compass);
  (void)north;
}

TouchSensor

• Types: bumper (boolean-like scalar), force (scalar), force-3d (vector).
• Use getValue() for scalar types, getValues() for force-3d.

Python:

touch = robot.getDevice("touch")
touch.enable(timestep)

while robot.step(timestep) != -1:
    if touch.getType() == touch.FORCE3D:
        fx, fy, fz = touch.getValues()
    else:
        f = touch.getValue()

C:

WbDeviceTag touch = wb_robot_get_device("touch");
wb_touch_sensor_enable(touch, time_step);

while (wb_robot_step(time_step) != -1) {
  WbTouchSensorType type = wb_touch_sensor_get_type(touch);
  if (type == WB_TOUCH_SENSOR_FORCE3D) {
    const double *f = wb_touch_sensor_get_values(touch);
    (void)f;
  } else {
    double f = wb_touch_sensor_get_value(touch);
    (void)f;
  }
}

RangeFinder

• Depth-camera style sensor.
• getRangeImage() returns per-pixel distance values.

Python:

rf = robot.getDevice("range_finder")
rf.enable(timestep)

width = rf.getWidth()
height = rf.getHeight()

while robot.step(timestep) != -1:
    depth = rf.getRangeImage()  # length = width * height
    center_depth = depth[(height // 2) * width + (width // 2)]

C:

WbDeviceTag rf = wb_robot_get_device("range_finder");
wb_range_finder_enable(rf, time_step);

int width = wb_range_finder_get_width(rf);
int height = wb_range_finder_get_height(rf);

while (wb_robot_step(time_step) != -1) {
  const float *depth = wb_range_finder_get_range_image(rf);
  float center = depth[(height / 2) * width + (width / 2)];
  (void)center;
}

Radar

• getTargets() returns detections.
• Target fields: distance, receivedPower, speed, azimuth.

Python:

radar = robot.getDevice("radar")
radar.enable(timestep)

while robot.step(timestep) != -1:
    targets = radar.getTargets()
    for t in targets:
        d = t.distance
        p = t.receivedPower
        s = t.speed
        a = t.azimuth

C:

WbDeviceTag radar = wb_robot_get_device("radar");
wb_radar_enable(radar, time_step);

while (wb_robot_step(time_step) != -1) {
  int n = wb_radar_get_number_of_targets(radar);
  const WbRadarTarget *targets = wb_radar_get_targets(radar);
  for (int i = 0; i < n; ++i) {
    double d = targets[i].distance;
    double p = targets[i].received_power;
    double s = targets[i].speed;
    double a = targets[i].azimuth;
    (void)d; (void)p; (void)s; (void)a;
  }
}

LightSensor

• getValue() returns irradiance/light intensity.

Python:

light = robot.getDevice("light")
light.enable(timestep)

while robot.step(timestep) != -1:
    irradiance = light.getValue()

C:

WbDeviceTag light = wb_robot_get_device("light");
wb_light_sensor_enable(light, time_step);

while (wb_robot_step(time_step) != -1) {
  double irradiance = wb_light_sensor_get_value(light);
  (void)irradiance;
}

PositionSensor

• Joint feedback sensor.
• Returns radians (rotational) or meters (linear).
• Attached to a Joint node.

Python:

pos = robot.getDevice("joint_sensor")
pos.enable(timestep)

while robot.step(timestep) != -1:
    q = pos.getValue()

C:

WbDeviceTag pos = wb_robot_get_device("joint_sensor");
wb_position_sensor_enable(pos, time_step);

while (wb_robot_step(time_step) != -1) {
  double q = wb_position_sensor_get_value(pos);
  (void)q;
}

Receiver

• Communication sensor paired with Emitter.
• Read packet queue with getQueueLength, getData, nextPacket.

Python:

rx = robot.getDevice("receiver")
rx.enable(timestep)

while robot.step(timestep) != -1:
    while rx.getQueueLength() > 0:
        data = rx.getData()
        # decode bytes as needed
        rx.nextPacket()

C:

WbDeviceTag rx = wb_robot_get_device("receiver");
wb_receiver_enable(rx, time_step);

while (wb_robot_step(time_step) != -1) {
  while (wb_receiver_get_queue_length(rx) > 0) {
    const void *data = wb_receiver_get_data(rx);
    int size = wb_receiver_get_data_size(rx);
    (void)data; (void)size;
    wb_receiver_next_packet(rx);
  }
}

Altimeter

• getAltitude() returns altitude above ground.

Python:

alt = robot.getDevice("altimeter")
alt.enable(timestep)

while robot.step(timestep) != -1:
    altitude = alt.getAltitude()

C:

WbDeviceTag alt = wb_robot_get_device("altimeter");
wb_altimeter_enable(alt, time_step);

while (wb_robot_step(time_step) != -1) {
  double altitude = wb_altimeter_get_value(alt);
  (void)altitude;
}

Cross-sensor integration patterns

• Fuse IMU (Accelerometer, Gyro, InertialUnit) for stable attitude estimation.
• Use Camera recognition + Lidar depth for robust obstacle classification.
• Use GPS + Compass for global heading and waypoint tracking.
• Gate expensive sensor reads by task phase (navigation vs manipulation vs idle).

Debug checklist

• Verify exact device name matches node name field.
• Verify enable(...) call executes before first getter call.
• Verify getter aligns with sensor type (getValue vs getValues).
• Validate sampling period and expected update rate.
• Confirm node fields (lookupTable, ranges, FOV, coordinate system) match design assumptions.

Reference file

Use references/sensors_reference.md for complete multi-language API signatures, node fields, data formats, and recognition object schema.