Raspi
Raspi is a library of light-weight services for programming Raspberry Pi’s.
The goal of the Raspi platform is to create an extensable, scalable, hardware automation platform capable of realtime evaluation of it’s state. The platform should be runnable on a single Raspberry Pi with exceptable performance, and scalable to a multiple Pi scenario sharing co-state and ‘work-together’ task orientation.
Some early design considerations
- The initial hardware platform is Pi3
- Using MegaPi industrial for DC motor drive
- Using GrovePi for instrumentation expansion
- ARMv7 Processor rev 4 (v7l)
- Each component of
Raspishould be self contained and plug-in replaceable/extensible. For lack or a better term, isolated/containerized. - intercommunication message serialization to start off as JSON, but will move to msgpack if necessary.
- Optimization should be on memory footprint
- Performance goal is 1.5x operational speed requirements, 1 second allowance for task response-time
- Python based for easy IO interaction with hardware components
Contents
Currently contains:
task_engine: Based on ZMQ, the primary orchestrator for handling both REQ => REP for
task automation, and PUB => SUB for sensor data retrieval.
In progress:
data_collector: Micro daemon collective for streaming sensor data to task_engine
subscribers.
data_store: Light-weight storage system for state samples and subscriber for
task_engine publishers.
hive_link: Interlink service to share state and operations between task_engines.
task_operator: Requestor for task execution, and interface for AI or remote-ops to
request task execution. Most likely based on Flask.
ai: High-level controller of sub_routines, such as pathfinding, and servo
feedback control. Primary client for task_operator.
tasks: Library of registered tasks, or system capabilities.
touch_ui: Interface for 7-inch touch-screen. Most likely based on TK.
monitoring: Data-visualization page for observing application.
Relationships
sub_routines
| ________ analysis
|/
| ________ motor_drive
|/
tasks (code_modules)
|
| ________ data_store
|/
ai _____> task_operator _____> task_engine [a] <_____ data_collector [Sensors]
/| ||
/ | ||
/ | hive_link
/ | ||
/ | ||
/ | task_engine [b, ...]
touch_screen ______> touch_ui |
|
|
|
|
|
browser ______> monitoring
Non-Core Libraries
pyzmq
Interaction Specification v0.1
| Service | Port | Notes |
|---|---|---|
| task_engine: workers | random-port | task_engine will assign a random port to workers since they register directly to the routers DEALER |
| router: frontend | 9000 | |
| router: backend | 9001 |