iRobot Roomba S9 vs Roomba J7 vs Roomba I7 vs Roomba 980 vs Roomba I3 - Navigation - Part I

Some basics about the navigation and some testing in an empty feature poor room. More testing soon...
Surface handling : • Roomba s9 s9+ Roomba j...

[@matteventu]

Excellent video as always!
It would be amazing if, in the obstacle avoidance part with j7 and s9, you could also compare the obstacle avoidance of Mijia 1T.
Also, another interesting test would be to see how much time the VSLAM robots take to understand what room/location they're in, if you lift them up and manually place them in a (previously mapped) location. On this, the Roomba and Dreame/Xiaomi (VSLAM) models have a completely different approach: Xiaomi/Dreame will first try to localise themselves moving in 4 directions in a "+" pattern (I guess to sort of "triangulate" their position based on the visual features they find) and then after they confirm their location they will start cleaning as per instructed cleaning (or, if they fail to understand the location, they'll just start a generic "clean all" task - after which usually they'll still be able to localise themselves after a few minutes or a few meters), while Roombas will immediately start cleaning and only after a while they'll understand the room they're in (visible in the app based on room name, as unfortunately they don't have a "live map" like other robots) and adjust the cleaning accordingly. Would be interesting to see what approach is better in feature-rich vs feature-poor rooms (also considering the VSLAM camera of Xiaomi/Dreame is not angled like that of Roombas and it directly points upwards).

[@frickhelm]

Oh yes, a good point i try to include maybe in a seperate video. Dyson is doing the same like the xiaomi / dreame despite its 360 degree view (i think this is due to the fisheye view and perspective erros coming with it). Judging from personal use (i use the s9 for my rooms) the roombas are very fast in relocating if started somewhere in a mapped area but as stated i will give it a test in a more controlled environment.

[@fritzmueller1337]

Bei den Roombas blickt doch keiner mehr durch wer jetzt was hat.
Wir haben einen i7 und es stört mich das der manchmal nicht mehr weiß wo er ist. Dann dreht er sich ein paar um die eigene Achse. Wollte mir dehalb den Roborock s7 maxv holen.
Kann man sagen, das sich die roborocks im allgemeinen besser orientieren ?
Danke für dein Video Frickhelm

[@frickhelm]

Pauschal kann man das so nicht sagen. Laserbasierte Roboter sind anfälliger für Umgebungsveränderungen im Startbereich und generell fehleranfälliger bei haushaltsüblichen Oberflächen wie Spiegeln, Glanzflächen etc. pp während die meisten kamerabasierenden Roboter nur eingeschränkt bei Dunkelheit nutzbar sind (Ausnahme bspw. Dyson Heurist 360). Bezüglich deines Problems würde mich interessieren wie da die Lichtverhältnisse sind und ob da bspw. Bilder hängen auf denen Personen zu sehen sind. iRobot hat nach einem Artikel in der washington post einen "Gesichter/Menschen" - Vermeidungsalgorithmus implementiert und der Roboter sucht nach neuen Blickwinkeln zur Orientierung, bei denen kein Mensch mehr im Bild ist.