1. Ceilbot locomotion by climbing arms and anchor points -The robot uses four extensible arms to grip metal anchor pins mounted in the ceiling. -Traveling across the room is achieved by changing anchor points on the way. Smaller position adjustment is done by extending and retracting the arms. - No rails or special materials needed, anchors are easily mounted. -Also works on a curved ceiling -The robot can avoid obstacles such as ceiling lamps easily. -It is possible to pass through doors if anchors are mounted on the wall above the door -Multi-axis positioning is possible, depending on the degrees of freedom for each arm.

2. Detail of the anchor pin and gripper arm concept A metal anchor bolt drilled into the ceiling The robot grips a pin shaped like a ball joint Each of the four telescopic arms has a gripping claw The arm can retract and extend telescopically, and it is attached to the robot by a 2-DOF joint.

3. Navigation in the operation environment using a multi-layered map -The ceilbot could use a map with multiple layers of complexity. -The different layers would store environment information based on how often it changes. -The highest layer would contain the extents of the room, i.e. the walls' and doors' positions, and other such data that is not likely to change. -Another layer would store the positions of, for example, pieces of furniture, which may change, but remain constant for a relatively long time. -Yet another layer would store the positions of individual objects which often change, for example books, dishes, shoes, chairs. -To navigate largely in the room, the robot then only needs to check if the higher maps still hold. This could be done by quick spot measurements, instead of continuously scanning and interpreting everything. -Temporary high resolution maps of certain parts of the room could be scanned when the robot needs to do something that requires precision. These maps wouldn't need to last a long time in memory, because they'd quickly expire when people move things around. -When first activated in a new environment, the robot would scan everything it encounters into the highest layer. As it operates in the same environment for a longer time, it would compare new scans with the old ones to average out frequently changing bits and move them to a lower layer. - Navigation could become progressively faster over time, as the robot could optimize its movements based on the mapping of fixed elements in its operation area.

4. Layered mapping concept sketch 1. An operating environment (a room in an apartment)‏ 2. A high-level map of the fixed elements in the environment (the walls and doorways)‏ 3. A mid-level map of the somewhat fixed elements in the environment (furniture etc.)‏ 4. The most often changing elements are stored in yet another mapping level. ( Suitcase on the floor, etc.) The knowledge of position of small objects in one end of the room is not necessary for operation in the other end, so high-res maps are obtained locally as needed. Gross navigation is based on global low-res maps.