Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 4 additions & 2 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -16,10 +16,12 @@ We have some zephyr rtos projects under zephyr_projects. To build/develop them y
### Arduino
We have some arduino projects, under arduino_firmware.

### common libraries
Under common_libs/ we have our common libraries. These should be hardware agnostic, modular and testable.
### Common Libraries
Under common_libraries/ we have our common libraries. These should be hardware agnostic, modular and testable.
We have an example here, that shows the general workflow for creating a common library module, including Catch2 testing.

1. [Low pass FIR filter](common_libraries/low_pass_fir_filter/README.md)

### Testing
We use Catch2 for testing cpp and c code. It's an easy to use, and easy to integrate tool for cpp unit tests.
Ideally, all our code has test coverage. Test driven development (TDD) is a powerful process where if done perfectly, you never push a bug that would impact system operations, because your tests would cover every needed operation of the system.
Expand Down
5 changes: 3 additions & 2 deletions common_libraries/CMakeLists.txt
Original file line number Diff line number Diff line change
Expand Up @@ -13,7 +13,7 @@ set(BUILD_DIR ${CMAKE_CURRENT_LIST_DIR}/build)

set(COMMON_LIB_SOURCES
${CMAKE_CURRENT_LIST_DIR}/example/some_lib.cpp

${CMAKE_CURRENT_LIST_DIR}/low_pass_fir_filter/low_pass_fir_filter.cpp
# Autogenerated files:
# ${BUILD_DIR}/autogen/*.cpp
# ${BUILD_DIR}/autogen/*.c
Expand All @@ -23,10 +23,11 @@ add_library(common_lib STATIC ${COMMON_LIB_SOURCES})

target_include_directories(common_lib PUBLIC
${CMAKE_CURRENT_LIST_DIR}/example/
${CMAKE_CURRENT_LIST_DIR}/low_pass_fir_filter/
)

add_executable(common_lib_tests
${CMAKE_CURRENT_LIST_DIR}/example/test/test_some_lib.cpp)
${CMAKE_CURRENT_LIST_DIR}/example/test/test_some_lib.cpp ${CMAKE_CURRENT_LIST_DIR}/low_pass_fir_filter/test/test_low_pass_fir_filter.cpp)


target_link_libraries(common_lib_tests PRIVATE
Expand Down
18 changes: 18 additions & 0 deletions common_libraries/low_pass_fir_filter/README.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,18 @@
# Low Pass FIR Filter
This is an FIR filter that can be used on any unsigned 32-bit integers. You can specify the order (AKA. the number of previous data points the algorithm considers) when creating an instance of LowPassFIRFilter (default order of 1). The following example filters incoming data with an order of 1:
```cpp
#include <low_pass_fir_filter.hpp>

LowPassFIRFilter filter;

void process_incoming_data(unsigned int input) {
unsigned int filtered_input = filter.update(input);
send_filtered_data_to_arm(filtered_input);
}
```

1. Constructor: defaults to order of 1
2. Buffer: used to store previous data points
3. Update function:
- Fills the entire buffer with the first input to simplify the algorithm (otherwise cases for when the buffer is not full yet have to be handled differently).
- Uses sliding window to calculate the average of the correct set of data points.
34 changes: 34 additions & 0 deletions common_libraries/low_pass_fir_filter/low_pass_fir_filter.cpp
Original file line number Diff line number Diff line change
@@ -0,0 +1,34 @@
#include "low_pass_fir_filter.hpp"
#include <stdexcept>
#include <cstdio>

LowPassFIRFilter::LowPassFIRFilter() : LowPassFIRFilter(1) {}

LowPassFIRFilter::LowPassFIRFilter(unsigned int order)
: buffer_is_empty(true), buffer(), buffer_index(0), order(order),
coefficient(1.0f / (order + 1)), output(0) {
// Valid input check
if (order < 1 || order > MAX_LP_FIR_ORDER) {
throw std::invalid_argument("order must be at least 1 and less than the max order");
}
}

unsigned int LowPassFIRFilter::update(unsigned int input) {
if (this->buffer_is_empty) {
// fill entire buffer with the first input
for (int i = 0; i < this->order + 1; i++) {
this->buffer[i] = input;
}
this->buffer_is_empty = false;
return this->output = input;
}

// `input` is the new data point that is begin added. `buffer[buffer_index]` is the old data point that must be removed.
this->output = this->output + input * this->coefficient - this->buffer[this->buffer_index] * this->coefficient;
this->buffer[this->buffer_index] = input;

// the index must wrap around once it reaches the right most side of the buffer. Note that buffer size is order + 1.
this->buffer_index = (this->buffer_index + 1) % (this->order + 1);

return this->output;
}
34 changes: 34 additions & 0 deletions common_libraries/low_pass_fir_filter/low_pass_fir_filter.hpp
Original file line number Diff line number Diff line change
@@ -0,0 +1,34 @@
#ifndef FIR_FILTER_HPP
#define FIR_FILTER_HPP
#define MAX_LP_FIR_ORDER (10)

/*
This class is used to apply a low pass FIR filter to a stream of inputs with a specific order. This filter is simply taking S=[ current_input, previous_input_1, previous_input_2, ... , previous_input_{order} ], and returning the average.

The buffer is initially filled with the first input to simplfy the algorithm (this removes the need for taking the average of less than "order" + 1 number of inputs).

The buffer is of size "order" + 1 since we need to store the current input, and also remember the oldest input (to remove from the total output in a subsequent update).
*/

class LowPassFIRFilter {
private:
bool buffer_is_empty;
unsigned int buffer[MAX_LP_FIR_ORDER];
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

AFAIK this would allocate more memory than needed if user does not use max order.
Great use case for a template...

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm used to prioritizing speed over space (static allocation reduces overhead a bit - compared to dynamically allocating vector at runtime), is space usually more important for firmware?

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

good question, but here a template will still be static allocation at compile time. I might be missing a use case so take this with a grain of salt, but regardless of a template or a hard coded buffer size its still up to the user to allocate the memory at runtime or compile time.
A vector also uses templates, but the template in a vector determines the type of data stored, not the size of the vector.

in my experience it depends on the project/team and micro-controller, but in robotics firmware i think speed is usually more important to prioritize. In other words, I've found its usually easier to allocate more memory without causing an issue compared to taking up more cpu time. Dynamic allocation on the heap is also just more prone to errors on a micro controller compared to a 'full computer' so always good to bias towards static allocation in firmware.

Sorry for the wordy explanation, happy to talk about this more over a call.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I had trouble creating a template that could be used across both hpp and cpp files. I left this out for now.

unsigned int buffer_index;
float coefficient;
float output;

public:
unsigned int order;
// Default constructor (order 1, equal weights)
LowPassFIRFilter();

// Constructor with filter order (uses equal weights)
// order = number of previous inputs to use
LowPassFIRFilter(unsigned int order);

// Update filter with new input and return filtered output
unsigned int update(unsigned int input);
};

#endif
Original file line number Diff line number Diff line change
@@ -0,0 +1,71 @@
#define CATCH_CONFIG_MAIN // Let catch2 handle the main function and boiler plate code.
#include <catch2/catch_all.hpp>
#include <low_pass_fir_filter.hpp>


Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Good usage of catch2.
Great test coverage.

TEST_CASE("test_constructor")
{

SECTION("test_no_order")
{
LowPassFIRFilter filter_instance;
REQUIRE(filter_instance.order == 1);
}

SECTION("valid_orders")
{
LowPassFIRFilter filter_instance_1(1);
REQUIRE(filter_instance_1.order == 1);

LowPassFIRFilter filter_instance_5(5);
REQUIRE(filter_instance_5.order == 5);

LowPassFIRFilter filter_instance_10(MAX_LP_FIR_ORDER);
REQUIRE(filter_instance_10.order == MAX_LP_FIR_ORDER);
}

SECTION("test_order_too_low_or_too_high")
{
REQUIRE_THROWS_AS(LowPassFIRFilter(0), std::invalid_argument);
REQUIRE_THROWS_AS(LowPassFIRFilter(MAX_LP_FIR_ORDER + 1), std::invalid_argument);
}

}

TEST_CASE("test_update")
{
SECTION("test_empty_buffer_same_input") {
LowPassFIRFilter filter_instance(5);
unsigned int val = 50;
for (unsigned int i = 0; i < 7; i++) {
REQUIRE(filter_instance.update(val) == 50);
}
}

/*
buffer = [50, 50, 50, 50, 50] -> output = 50
buffer = [100, 50, 50, 50, 50] -> output = 60
buffer = [100, 20, 50, 50, 50] -> output = 54
*/
SECTION("test_empty_buffer_different_inputs") {
LowPassFIRFilter filter_instance(4);
unsigned int inputs[] = {50, 50, 50, 100, 20};
unsigned int expected[] = {50, 50, 50, 60, 54};
unsigned int n = 5;
for (unsigned int i = 0; i < n; i++) {
REQUIRE(filter_instance.update(inputs[i]) == expected[i]);
}
}


SECTION("test_input_spikes") {
LowPassFIRFilter filter_instance(1);
unsigned int inputs[] = {50, 60, 70, 500, 100, 60, 50, 40, 50, 40, 50};
unsigned int expected[] = {50, 55, 65, 285, 300, 80, 55, 45, 45, 45, 45};

unsigned int n = 11;
for (unsigned int i = 0; i < n; i++) {
REQUIRE(filter_instance.update(inputs[i]) == expected[i]);
}
}
}