State-Machine Based Decimation Filter for LiDAR UART Streaming
G
Gian Fajardo, SPaRA ResearchA state machine for reading a continuous stream of 5-byte UART messages from a LiDAR scanner into an embedded system — in real time, with configurable decimation. Rather than capturing everything and filtering later, this design processes bytes as they arrive and skips every Mth message on the fly. Five states handle the full lifecycle: idle, pattern detection, offset alignment, recording, and skipping. Built for embedded systems engineers, robotics researchers, and anyone working with sensor data pipelines where memory is constrained and throughput matters.
How to create a State-Machine Based Decimation Filter for LiDAR UART Streaming
To create a state-machine based decimation filter for LiDAR UART streaming, follow these steps:
01.
Define your states before your transitions
List every mode the system can be in: HOLD (idle), FIND_PATTERN (sync detection), ADD_OFFSET (byte alignment), RECORD (capture), SKIP (discard). Each state has exactly one job.
02.
Assign a limit to each state
Every state runs until a counter reaches a state-specific limit — WAIT_INDEX for HOLD, FIND_INDEX for FIND_PATTERN, offset for ADD_OFFSET, MSG_LENGTH for RECORD, SKIP_INDEX for SKIP. Document these as notes on each state node.
03.
Model the init transition
Start from [*] and arrow into HOLD — this represents system initialization and makes the entry point explicit.
04.
Use a choice node for branching logic
After FIND_PATTERN detects a candidate pattern, there are two possible next states depending on whether an offset correction is needed. A <<choice>> node makes this branch readable without cluttering the state definitions.
05.
Close the decimation loop
RECORD → SKIP → RECORD is the core cycle. The counter inside each state determines how many bytes get captured vs. discarded, implementing the 1-in-M message decimation.
06.
Add the buffer-full exit
SKIP → HOLD fires when interm_buffer_counter > PROCESS_BUFFER_SIZE — the system has filled its output buffer and resets to wait for the next read cycle. Model this as a guarded transition with the condition labeled on the arrow.
07.
Use notes to carry implementation details
State diagrams communicate structure, not values. Use note left of and note right of to attach the specific limit constants to each state without cluttering the transition arrows.
You might also like
View all View all templatesMaterial Tracking ERD
A detailed entity-relationship model for a material hauling and tracking system. Covers customers, jobs, tickets, materials, haulers, and billing — with all foreign key relationships and field-level detail. A solid starting point for teams building dispatch, logistics, or field service software who need to define their data model before writing schema migrations.
R
Renso Höllhumer, Solutions Architect
Customer & Order ER Diagram
Visualize how customers, orders, invoices, products, and delivery addresses relate in your system. This ER diagram helps clarify relationships, responsibilities, and dependencies at a glance, supporting database design and system documentation.
M
Mermaid
C4 Context Diagram
Show the big picture of how your system fits into its environment using the C4 model approach. This template maps users, your system, and external dependencies with clear boundaries — perfect for explaining system scope to stakeholders, planning integrations, documenting architecture decisions, or onboarding new team members to complex platforms.
M
Mermaid
UML Class Messaging System
Map the structure of a messaging system from the ground up. This template lays out core entities — conversations, messages, users/contacts, channels, attachments, and notifications — plus how they relate. It’s a solid starting point for API design, data modeling, and implementation planning.
M