IntermediateTopic 68 of 102Manufacturing, Robotics & Systems

Sensor fusion

Combine IMU, GPS, lidar and cameras for robust estimation.

In depth

Sensor fusion combines complementary sensors for better estimates: complementary filters, Kalman/EKF/UKF, particle filters and factor graphs. Examples: IMU+GPS+wheel odometry, camera+lidar+radar in vehicles. Needs time synchronisation, calibration and noise models.

Key points

  • Covariance weighting
  • Redundancy for fault detection
  • Latency compensation
Example

Drones fuse gyro, accelerometer and barometer to stabilise attitude and altitude.

Why it matters

Sensor fusion combines several sensors so that the weakness of one is covered by another, which gives more accurate and reliable estimates.

Where this is used

  • Combining GPS with inertial sensors for a vehicle position.
  • Fusing lidar and camera data for obstacle detection on a robot.
  • Estimating the attitude of a drone from gyroscope and accelerometer data.

Frequently asked questions

What is Sensor fusion?

Combine IMU, GPS, lidar and cameras for robust estimation.

Why is it important to understand sensor fusion?

Sensor fusion combines several sensors so that the weakness of one is covered by another, which gives more accurate and reliable estimates.

Why combine sensors?

Each sensor has errors and blind spots, so combining them can give better accuracy and robustness than any one alone.

What is an IMU?

An inertial measurement unit with accelerometers and gyroscopes, which measure motion and orientation but drift over time.

How is Sensor fusion used in practice?

Typical applications include combining GPS with inertial sensors for a vehicle position, fusing lidar and camera data for obstacle detection on a robot and estimating the attitude of a drone from gyroscope and accelerometer data.

What should I learn before Sensor fusion?

Study "Path planning" first (topic 67 of 102 in Manufacturing, Robotics & Systems). It covers the groundwork this topic builds on.

What are the key points of Sensor fusion?

Covariance weighting. Redundancy for fault detection. Latency compensation.

Can you give an example of Sensor fusion?

Drones fuse gyro, accelerometer and barometer to stabilise attitude and altitude.

What comes after Sensor fusion?

The next topic is "ROS", which builds directly on this one.

How difficult is Sensor fusion?

It is classed as Intermediate in Manufacturing, Robotics & Systems (topic 68 of 102). It is an intermediate topic. Work through the beginner topics in Manufacturing, Robotics & Systems first.

Learning path

Want to practise? Open the interactive course for quick checks, quizzes, simulations and progress tracking.

Related topics