Clean it is an automatic vaccum cleaner that has an autonomous mode that mimics the algorithm used by Roomba for ensuring maximum coverage of the space. It also has 4 preprogramed paths to follow:
-
Empty room: move straight until colliding with wall, upon collision it turns aroud 180° while slightly moving to the side and move straight until colliding again. This will repeat until timer is done.
-
Spiral movement
-
Baterry Test: this is a system validation tool, it simply sets the motors to full capacity.
-
Room Outline: in this mode, Clean It will detect a wall to it's right and follow along it's perimeter. It will also turn on corners.
Both the autonomous mode as well as each of th patterns can be set to run for a certain amount of time, which can be inputed using the keypad on the lid. Additionally, the patterns have a Deep Clean mode in which Clean It will stop between movements to ensure it will stay in place longer.
Clean It has the follow key componets that make up it's functionality
-
Limit Switches: used to detect collisions with wall.
-
Proximity sensors: infrared sensor that output variable voltage depending on the distance.
-
H-Bridge and motors: forms the drive train of the system, uses PWM to control vaccum at different speeds.
-
Keypad and LCD: used for IO.
-
Turbine: a turbine is needed to create the preassure difference necessary to displace air in the intake.
-
Power Distribution: Clean It is powered by 3 18650 cells, however 5V are necessary for many of the above components including the microprocessor, thus a 5V regulator was implemented and screw terminals where used to map all the power lines.
-
KL25Z Board and GPIO Board: the brains of this project is NXP's KL25Z Freedom Board. To interface the board with the components screw terminals where routed to the GPIO pins using a "PCB" of sorts, ensuring a robust system.
Clean It was built using the KL25Z Freedom Board. The code is structure in layers of abstraction, first including simple headers to interface the sensors and actuators and then processing this data to understand Clean It's surroundings. The end goal for the sensor abstractions was to have functions that detect collisions (using limit switches) and posible collisions (using proximits) in boolean terms. As for the actuators, the movements of the vacumm where expressed in terms of directio and parametrized using timers.
In managing this layered approach, the final driver code for the vaccum is much less complex, and it also encourages further contribution and personalization from the community.
| Abstraction Level | Description |
|---|---|
| Hardware level | The necessary registers are configured for each sensor and actuator, forming an init function for each component. The actual reading and/or writting to the GPIO pins also occurs here. |
| Data in context | Once the sensors have been interfaces, the data is read and used to trigger flags depending on the value. For example, the proximity sensors and limit sensor work together to form the collision detection system. |
| Route Planning | Finally the processed data can be used to encapsulate the routines the vaccum will perform |