Just two items I did not cover.
1. The software handling error conditions, blocked wheels etc.
When the Powerbot stops on an error C01, C02 etc., it cannot be resumed to complete the current cleaning operation, and must be restarted from the beginning. This is not true of other brands. They can show "recovering position", and resuming the plan of a house. There are many considerations involved. If the robot is unattended, leaving the computer running for a day can exhaust the battery, causing damage without proper power management software. Powerbots are programmed to return to base automatically after half an hour if not cleaning, though I am not sure on an error condition.
Nevertheless, consider cleaning large spaces where two or three charging cycles may be used to clean the entire house automatically. Starting over is then not as simple as doing just a single charge run.
Ideally it should be possible to retain the current cleaning operation map and information so as to resume after clearing a simple error such as a wheel stall. I had something like that just from the orbot maneuvering in a tight space, which is usually handled, but some rare positioning developed in that one instance.
The larger size and speed of the Powerbot makes it practical to run the robot attended, and it can even be entertaining. If the robot gets stuck somewhere it would be an important convenience to be able to release the bot from its trap and resume the operation. Users would credit Samsung with that kind of detail.
2. Tight space maneuvering
All brands seem to suffer from lack of intelligence navigating in tight spaces, such as under a cabinet between furniture legs, or between furniture and walls etc. The 9000 Powerbot requires about 20 inches of space in which to maneuver, understandable as the front must swing around in order to turn.
However, when entering a tight space robots seem to employ a random sort of procedure bumbing around and turning incessantly in order to get out. They seem to have no memory of how they got into a tight space in the first place. With newer models showing maps of the robot movements through spaces, one would think this information could be used for a smarter procedure. Processing the path by which a robot enters a space ought to indicate the right direction for getting out of the space expeditiously. Robots act as though they have no clue.
The Powerbot has performed better on this than others I have had, not getting permanently trapped, but still show the dumb kind of procedure -- just implemented more correctly perhaps.
A request came to my attention, if the home automation detects the resident has come home, automatically instruct vacuum robots to terminate cleaning and return to base, for convenience. Maybe some extension of the scheduling function.
Suggestion for future development of "Select a room" and boundary marking. A frequent request or complaint about vacuum robots is when the robot does not finish one room at a time, and goes back and forth between rooms. Doing whole rooms in order helps knowing where the robot is and likely will be over time, when the resident is home while cleaning.
For efficiency minimizing time consuming turns, the Powerbot I have likes to travel as far as possible in one direction, even through doorways when so aligned. The style makes an impression of efficiency, and less seemingly random maneuvers, attractive.
The Virtual Guard optical boundary system inherited from the old Navibot series, replaced on newer models only in the U.S. with magnetic strip markers similar to Neato Robotics, never worked well for lack of focus in the light beam, reflections off walls, etc. The Virtual Guard was never expanded to have iRobot's "lighthouse" function of passing a robot from room to room, implementing room completion one at a time. I suspect the poor performance of the original, old Guards and the plan to supply mag strips instead terminated any expansion of the Guard.
The mag strips, however, do not implement the "lighthouse" function, alternating between barrier mode and passing mode as needed (which decision has to be made inside the robot).
The future is in the "persistent mapping", remembering the house map between runs supporting the "select a room" feature. Other makers are now finally introducing marking such maps on the smartphone with virtual boundary lines to replace physical or electronic barriers. As Samsung develops its own further features for persistent mapping, using that ability to implement room completion could prove attractive to some buyers.
Perhaps one "virtual boundary"in a room could be tagged as a passage between rooms to guide the operations. Even were camera vision able to identify standard doorways, they would not work on various wider openings etc., so some user input is needed.
Actually this could be just a simple extension of the "select a room" feature: add a sequence of rooms to automatically select one at a time, covering the whole floor plan with one command. It might be possible just in the app code, or even in the Alexa intergration code (which can be user supplied, given the Amazon support available for DIY "skills").
Thank you. Any update from Samsung regarding the tight space issue? My 9350 keeps getting stuck under the dining room chairs. I assume this should be a relatively simple software/logic fix. Samsung, please respond and fix. Thank you.
Complete details for wheel mod at robotreviews.com Samsung forum, thread "PowerBot Side Bumper Extension and Tilt Limiting Mods", for the 9000 series but you can adapt to the 7000 series. The 7000 has less need for the side bumper extension. [url] http://www.robotreviews.com/chat/viewtopic.php?f=22&t=19388[/url]
If your powerbot does not support mag strip boundary markers, it is possible to add this with hobby electronics work. See thread in the "hacking" forum at the above website, "Powerbot mag strip addition mod" -- may be relevant more to European models where Virtual Guards were still sold, compared to American models with mag strips.
For the wheels a bit of cardboard or wood is fastened into the end of the wheel well to hold in the tip of the wheel suspension arm, with a lot of tape to resist the spring pressure -- or a simpler thumb tack into the plastic. The wheel must be held in enough to prevent tripping the switch sensor for lifting the robot, but leaving room for some tilting for climbing etc.