Sensor Fusion Engineer

Notes from the 2020 course

LiDAR

LiDAR (Light Detection And Ranging) is a sensor that measures position and distance (like RADAR) by using light (usually in the infrared spectrum). It is highly expensive (especially the rotating ones) but offers certain advantages over RADAR and helps augment missing data that the other sensors might not be able to capture (e.g. dark photo for a camera).

Camera

Cameras are abundant and cheap sensors that capture color data of an environment. The data is unfortunately flat, but can be augmented with computer vision techniques. Benefits to using it over lidar and RADAR is that color data can better classify objects.

Stereocameras exist but they are more expensive and require proper calibration.

RADAR

RADAR has been used for many years and is cheap compared to LiDAR while providing depth data. It can capture object velocities using the Doppler effect making it useful to augment with LiDAR. It doesn't work well with various reflective surfaces and is not affected by snow or other issues that LiDAR comes across.

Kalman Filters

The Kalman Filter is an algorithms that uses matrices to combine information from multiple sensors and obtain more accurate readouts over time. It has an averaging effect that can remove noise.

Course Opinions

TBD