
019c5c8a87ba9963ea5b18a9bdf6a7d4.ppt
- Количество слайдов: 19
Juhana Leiwo – Marco Torti CEILBOT PROJECT Final Presentation
Environments House Hospital Industry Warehouse Sports Hall Airport Restaurant Condominium SMS sending for important tasks execution and emergencies.
First hypothesis of structure
First hypothesis of structure
Focusing on MAPPING Multilevel map is the priority 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) 4. The most often changing elements are stored in yet another mapping level. (Suitcase on the floor, glasses, books, etc. ) 3. A mid-level map of the somewhat fixed elements in the environment (furniture etc. )
Useful data Position and movement Direction of acceleration (gravity) Proximity and collision sensing 3 -dimensional spatial mapping Camera vision Heat detection (to recognize humans, for example) Sound sensing (for voice commands and human activity sensing)
Price & performances The price of the robot is critical because the mapping feature is important in houses too For a multifunctional robot it could be useful to determine a number of basic senses and optional special senses (and so features) Multiflow data fusion with Kalman filter (or similar)
Mapping guidelines The robot starts making its own maps automatically when it gets installed in a new environment. It doesn't need preprogrammed maps The map gets continuously updated when the robot makes new measurements while operating
Mapping guidelines The mapping data is stored in different "layers" based on whether the measured area has changed. There are four different layers: 1. the high-level map, which contains the unchanging features of the environment 2. the medium-level map, which contains things that have changed but still exist 3. the change map, which describes how often different areas have changed 4. the updating layer, which contains the results of the newest scan.
Mapping guidelines The whole highest layer is used as a landmark when calculating location. Since the robot operates in the same environment, the map becomes more reliable over time as it is updated. A full scan is not always necessary for locating. Smaller point scans can be used instead The data in the medium-level map and the change map can be used for route planning and optimizing tasks like seeking an object from a room
Focusing in the ceil-walking robot
2 DOF joints Telescopic walking arms Gripping claws IR proximity sensors under the infrared transparent covering Part details Camera on rotating mount Manipulator arm
New advanced design Very complex but able to join all the strengths of the several designs
Sensors selection – Basic Sensors ACTIVE CURB FEELER (WISKERS) SYNCRO FORCE SENSORS LASER RANGE CAMERA (PSM) or STRUCTURED LIGHT
Sensors selection – Optional Sensors TIME OF FLIGHT WITH LASER ACCELERATION SENSOR INFRARED CAMERA THERMOGRAPHIC CAMERA
Security Stability – At least 2 legs attacked Wiskers – Hurt prevention Vocal keyword & Emergency button – To stop the robot Max weight Living & objects
Mapping algorithm guidelines S L A L O Scan imultaneous ocation Move and track with blind senses Scan again Match scans nd Update location Average old and new maps ayered ptimized Update high-level map Find points that would have been visible in both maps but only exist in one of them Update medium-level map apping Update change frequency map
Alimentation The main alimentation of the robot will be inside the gripping hooks The robot will also need a little battery inside (few minutes autonomy) to move where the alimentation is not available
Conclusions The project is very interesting and may have a use in a lot of environments The selection of the sensors is the critical part to obtain a good quality/price ratio The development of the mapping algorithm could be long and hard but probably will be one of the reasons that could determine the success of this product
019c5c8a87ba9963ea5b18a9bdf6a7d4.ppt