A canonical URL declares which version of a page is authoritative when the same content is reachable at several addresses. It consolidates ranking signals onto one URL while leaving the duplicates accessible to visitors.
Most common causes of duplication: tracking parameters, trailing slashes, HTTP and HTTPS, and www versus non-www.
Self-referencing canonicals on every page are good practice, not redundancy.
A canonical is a hint, not a directive — search systems can ignore it if the signal conflicts with everything else.