Glossary

Fictionlab > Glossary > Global shutter

Global shutter

Global shutter – definition

A global shutter is an image-sensor exposure method in which all pixels in a frame begin and end exposure at the same time. The sensor then reads the captured image data after the common exposure interval. This differs from a rolling shutter, where rows are exposed and read sequentially over time.

In mobile robotics and unmanned ground vehicles (UGVs), global shutter cameras reduce geometric distortion caused by robot motion, wheel vibration, fast turns, or moving objects. A frame captured during motion represents one exposure instant more consistently than a rolling-shutter frame. This is important for visual odometry, visual-inertial odometry, feature tracking, stereo perception, fiducial detection, and camera-based navigation.

Global shutter does not make a camera immune to motion blur. Blur still occurs when the exposure time is too long relative to image motion. It also does not replace camera calibration, timestamping, or synchronization with an IMU, LiDAR, or wheel odometry source.

Global shutter versus rolling shutter in a UGV

The practical difference is temporal. A rolling-shutter sensor captures different image rows at different moments. If a UGV rotates or travels over uneven terrain during frame acquisition, straight lines may appear tilted or bent, and tracked features may move inconsistently between rows.

A global-shutter sensor captures the full image during a shared exposure period. This improves the validity of the common assumption used by many perception pipelines: one image corresponds to one camera pose.

  • Global shutter: all pixels integrate light during the same exposure window.
  • Rolling shutter: pixel rows integrate light in sequence, with a row-dependent capture time.
  • Motion blur: can affect both shutter types when exposure is long.
  • Low-light trade-off: a short exposure reduces blur but may require more illumination, a wider aperture, or higher sensor gain.
  • Synchronization: external trigger support can align a global-shutter camera with an IMU, strobe, or another camera.

For a differential-drive platform, rotation around the vertical axis is a common source of image distortion. During in-place rotation, the apparent motion of scene features can be large even when the robot does not translate. Global shutter is therefore particularly useful for cameras mounted on Leo Rover or Raph Rover when the camera stream is used for state estimation or geometric perception.

Key parameters for global-shutter cameras

Shutter type alone is not enough to select a camera for autonomous robotics. The required parameters depend on the vehicle speed, vibration level, lighting, lens field of view, compute capacity, and the perception algorithm.

Parameter Why it matters for UGV perception
Exposure time Controls motion blur and collected light. It should be short enough for the expected image motion.
Frame rate Defines the interval between measurements and affects feature displacement between frames.
Hardware trigger Allows deterministic capture timing and multi-sensor synchronization where supported by the camera.
Timestamp source Determines whether software timestamps represent reception time or a camera-generated capture time.
Resolution, sensor size, and pixel size Influence feature detail, field coverage, low-light performance, bandwidth, and compute load.
Lens and calibration model Determine geometric accuracy. The camera model must match the calibrated lens.

A useful engineering constraint is the image displacement during exposure. For a projected feature velocity vpx in pixels per second and exposure time texp, approximate blur length is:

blur_px ≈ v_px × t_exp

This estimate does not include lens effects, vibration, or depth variation, but it shows why a global shutter and a suitably short exposure are complementary requirements.

Global shutter in ROS 2 camera pipelines

ROS 2 does not define a separate message type for global-shutter images. The shutter architecture is a property of the physical camera and its driver. A camera driver normally publishes image data as sensor_msgs/msg/Image and intrinsic calibration data as sensor_msgs/msg/CameraInfo.

The image header timestamp is operationally important. For visual-inertial pipelines, it should represent the image acquisition time as closely as the hardware and driver permit. A host-generated timestamp may include USB, Ethernet, operating-system scheduling, and driver latency. This can create time offsets relative to IMU measurements.

ros2 topic list | grep camera
ros2 topic echo --once /camera/image_raw --field header
ros2 topic echo --once /camera/camera_info

The frame_id in image and camera information messages should identify the optical sensor frame. Transform conventions should follow REP 103, while the relationship between base_link, odometry, and map frames is commonly structured according to REP 105. The optical frame orientation and the static transform to the rover body must be defined consistently in URDF and TF2.

Use with Leo Rover and Raph Rover

Leo Rover is a four-wheel, differential-drive platform with an integrated Raspberry Pi compute unit and ROS 2 support. A global-shutter USB camera can be integrated as a sensor payload when its Linux driver, ROS 2 driver, bandwidth requirements, power requirements, and mounting arrangement have been verified. It does not provide autonomous navigation by itself. A navigation stack, localization method, collision sensing, and application-specific safety logic are still required.

On Leo Rover, a global-shutter camera is suitable for laboratory work involving AprilTag detection, visual odometry experiments, line or terrain feature tracking, and camera-IMU synchronization research. Compute load should be measured on the installed Raspberry Pi, especially when image rectification, neural inference, recording, and SLAM run concurrently.

Raph Rover provides a larger payload-oriented UGV platform. It can accommodate more substantial sensor assemblies, including protected camera enclosures, synchronized stereo rigs, dedicated compute hardware, or combined LiDAR-camera payloads. Global shutter is relevant when those sensors support mapping, inspection, agricultural observation, or construction-site perception under vehicle motion.

Limitations and integration checks

A global-shutter camera should be evaluated as part of a complete sensing chain. The following checks identify common integration failures before field deployment.

  • Verify that the driver publishes correct image encoding, dimensions, camera calibration, and timestamps.
  • Measure camera-to-IMU time offset when using visual-inertial odometry.
  • Use rigid mounting and inspect images for vibration-induced blur.
  • Protect lenses from dust, water droplets, and direct glare in outdoor deployments.
  • Record ROS 2 bags during representative rover motion and validate feature tracking after sharp turns.
  • Confirm that exposure, gain, and white-balance settings do not change unpredictably during experiments.

Normative references and technical sources

Global shutter is a sensor architecture rather than a ROS standard. ROS defines message interfaces and coordinate conventions, while camera manufacturers define sensor timing, triggering, exposure, and interface characteristics in device specifications.

See also